joined by PHILLIPS, Chief Justice, GONZALEZ and OWEN, Justices, concurring in the judgment.
The result in the present case is not a hard call. The Court is unanimous. The issue is how the result is determined in this and other insurance bad-faith cases. On that subject the Court is deeply divided.
Ever since the motion for rehearing of the denial of the petition for writ of error in State Farm Fire & Casualty Co. v. Simmons was filed three and one-half years ago, the Court has searched for a way to define the common-law standard of bad-faith insurance liability more precisely and reconcile it with Texas’ traditional standard of no-evidence review. Bad-faith liability, which this Court created, depends in part on a determination that an insurer had no reasonable basis for denying or delaying payment of a claim. The reasonableness of the insurer’s decision can be judged only by weighing the evidence for and against the claim. But no-evidence review of a jury finding forbids weighing of evidence by an appellate court. The court is permitted to consider only undisputed evidence and evidence supportive of the finding. Thus, by definition, no-evidence review of a jury finding of bad faith based on conflicting evidence is impossible. Excluding such review deprives courts of their role in defining bad-faith liability, making bad faith whatever any particular jury thinks it is. Moreover, creating a cause of action that cannot be fully reviewed on appeal thwarts appellate courts from discharging their obligation to exercise the jurisdiction constitutionally assigned them, which includes evidentiary review.
This is the dilemma the Court faces, and it is one of the Court’s own making. Having created bad-faith liability, the Court ought to be able to define it and provide for meaningful evidentiary review. But as one court of appeals has candidly observed, “the Supreme Court, in its attempt to clarify the legal sufficiency standard for bad-faith claims against insurance companies, has ultimately done little to provide lower courts with any guidance for conducting a legal sufficiency review.” Columbia Universal Life Ins. Co. v. Miles, 923 S.W.2d 803, 810 (Tex.App.—El Paso 1996, writ pending). The failure is not for want of trying. For three and one-half years the Court has held almost every case filed involving an evidentiary challenge to a judgment for insurance bad faith. At least ten such cases are now pending. Besides this case, they are: State Farm Fire & Cas. Co. v. Simmons, 857 S.W.2d 126 (Tex.App.—Beaumont 1993), pet. for writ of error denied, 37 Tex. Sup.Ct. J. 301 (No. D-4095, Jan. 5, 1994), reh’g granted, 40 Tex. Sup.Ct. J. 930, — S.W.2d -, 1997 WL 378632 (July 9, 1997); Nicolau v. State Farm Lloyds, 869 S.W.2d 543 (Tex.App.—Corpus Christi 1993), pet. for writ of error granted, 37 Tex. Sup.Ct. J. 951 (No. 94-0287, June 15, 1994), aff'd, 951 S.W.2d 444 (Tex.1997); State Farm Lloyd’s Ins. Co. v. Ashby AAA Automotive Supply Co., No. 05-92-01354-CV, 1995 WL 513363 (Tex.App.—Dallas 1995), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 305 (No. 96-0120, Feb. 9, 1996); St. Paul Surplus Lines Ins. Co. v. Dal-Worth Tank Co., 917 S.W.2d 29 (Tex.App.—Amarillo 1995), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 317 (No. 96-0148, Feb. 16, 1996); Williams v. Crum & Forster Commercial Ins., United States Fire Ins. Co., 915 S.W.2d 39 (Tex.App.—Dallas 1995), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 383 (No. 96-0237, Mar. 13, 1996); Provident Am. Ins. Co. v. Castaneda, 914 S.W.2d 273 (Tex.App.—El Paso 1996), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 439 (No. 96-0249, Mar. 15, 1996); British Am. Ins. Co. v. Lewis, (Tex.App.—Waco 1996), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 447 (No. 96-0297, Apr. 2, 1996); Columbia Universal Life Ins. Co. v. Miles, 923 S.W.2d 803 (Tex.App.—El Paso 1996), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 1099 (No. 96-0822, Sept. 5, 1996); St. Paul Ins. Co. v. Escalante, No. 07-94-00214-CV (Tex.App.—Amarillo 1996), pet. for writ of error filed, 39 Tex. Sup.Ct. J. 1100 (No. 96-0794, Aug. 26, 1996); State Farm Lloyds Ins. Co. v. Maldonado, 935 S.W.2d 805 (Tex.App.—San Antonio 1996) (en banc), pet. for writ of error filed, 40 Tex. Sup.Ct. J. 136 (No. 96-1179, Dec. 2, 1996). I am not aware of another time when so many cases awaited resolution of a single issue.
The Court’s failure to define bad faith and provide for no-evidence review is for want of *59will. The cases pending here and dozens more in other jurisdictions demonstrate that a viable cause of action for insurance bad faith must have, as an absolute necessity, a definite legal standard of liability. This standard the Court unfortunately refuses to provide. Replacing the “no reasonable basis” standard of common-law bad-faith insurance liability with the more recently enacted “reasonably clear” statutory bad-faith standard simplifies the law but does nothing to clarify what bad faith means or resolve the dilemma of evidentiary review. To the claim in Justice Speotor’s opinion that merely switching words achieves something more than a relatively trivial unification of common-law and statutory standards, I say, “Show me the case.” Name one bad-faith case decided today, or one that remains pending, or just hypothesize one that would be reviewed or decided any differently by any judge under either the “no reasonable basis” standard or the “reasonably clear” standard. None exists. No JUSTICE’S view of a single case has been altered one iota by the change in standard. Justice Enooh is exactly right about one thing: all these bad-faith cases could have been decided years ago with the same results, the same confusion, and a lot less delay. The words mean nothing. The change in terminology is purely semantic, nothing more.
But JUSTICE Enoch is not correct in arguing that bad-faith law is fine as it is and no change is needed. He alone on the Court has this view, and he cites no one else who shares it. The overwhelming consensus of scholarly commentary, which I will examine in some detail, is that bad faith remains a vague, changing concept, that the tort has not “matured” into a cause of action with known, predictable parameters, and that to be viable it must be made more definite. These things are true in Texas and in all jurisdictions that recognize a bad-faith tort.
Justice SpectoR’s vaunting of an altered standard as a solution to the bad-faith dilemma is betrayed by her alternative position, advanced with equal force and straight face, that no change is really necessary because no significant damages can be recovered for bad faith. Bad-faith liability is no longer a problem, Justice Spector’s opinion assures, because the Court has tightened the standards for awarding mental anguish and punitive damages, and these are most of the damages in bad-faith cases. If bad faith really afforded no significant recovery of damages, one could wonder why the cause of action should exist at all, but one’s effort would be better spent trying to understand what Justice Speotor thinks is significant. Simmons involves awards of $200,000 in mental anguish and $2 million in punitive damages on a policy claim of $75,000. Dalr-Worth involves an $11.5 million punitive damages award on a policy claim of several thousand dollars. Even in today’s “enlightened” view of damage awards, such sums are hardly insignificant.
The essence of JUSTICE Spector’s view is revealed in the pattern of today’s decisions: no verdict of bad faith is overturned for lack of evidence. The jury has become fact finder and final arbiter, subject only to the court of appeals’ review of the factual sufficiency of the evidence. Unable to define bad faith, Justice Spector’s opinion would cede that responsibility to juries who must now do what the Court refuses to do. To draw attention from this fact, Justice Spector’s opinion alleges that I “would take the resolution of bad-faith disputes away from the juries.” Confident as I am — and remain — of juries’ ability to determine credibility and find facts, I think it is most unfair to expect them to determine fact and policy — to enact, find, and hold, all three. A juror who believes insurers should err on the side of paying a claim will think the failure to do so unreasonable. A juror who believes insurers should be stricter in adjusting claims will be less likely to find denial of a claim unreasonable. Who is right? It is one thing to ask a jury whether proscribed behavior has occurred; it is quite another to ask whether the behavior that has occurred should be proscribed.
I would define bad faith as unscrupulous, arbitrary conduct and make the standard one of law as it is in many other instances, such as: whether an action under the Deceptive Trade Practices Act was groundless and brought in bad faith; whether a writing is *60ambiguous; whether a provision is conspicuous in a document; whether a person had probable cause for prosecuting charges; and whether language is defamatory. I would impose liability for intentional or reckless conduct on the part of an insurer but not merely for negligence or mistakes.
I
A
Since about the turn of the century courts throughout the country have imposed a duty on insurers to consider their insureds’ interests in settling liability claims against their insureds. Stephen S. Ashley, Bad Faith Actions, Liability and Damages §§.2.02-.07 (1992). Texas did so in G. A. Stowers Furniture Co. v. American Indemnity Co., 15 S.W.2d 544 (Tex. Comm’n App.1929, holding approved). An action for breach of the Stow-ers duty affords the remedy of extra-contractual damages to an insured who suffers a judgment in excess of coverage because of the insurer’s wrongful failure to accept a settlement offer within policy limits. Absent such a duty, an insurer has little incentive to settle a claim against its insured because its risk is limited to the amount of coverage. Ashley, supra, at § 2.02. Breach of the duty is usually characterized as a tort — either negligence, as in Texas, Maryland Insurance Co. v. Head Industrial Coatings and Services, Inc., 938 S.W.2d 27, 28 (Tex.1996) (per curiam), or bad faith. Ashley, supra, at §§ 2.03-.05. Because the duty arises in the context of a claim by a third party against the insured, the cases are usually referred to as third-party cases. Although important differences in the breadth of the duty have emerged among American jurisdictions, id. at 2.06, liability in third-party cases in this State and elsewhere is “a relatively mature, stable, and complete body of law”, id. at § 5.01.
So-called first-party eases involve claims by insureds for policy benefits for their own damages rather than indemnity for injuries to third parties. Courts have been more reluctant to impose a tort duty on insurers to settle first-party claims. For one thing, an insurer’s and an insured’s interests are not aligned when the insured is claiming on his own behalf as they are or should be in third-party cases where insurer and insured face a common opponent. Douglas R. Richmond, An Overview of Insurance Bad Faith Law and Litigation, 25 Seton Hall L.Rev. 74, 104-105 (1994) [hereinafter Richmond I]. While insurers are obliged to pay valid claims promptly, they are entitled to challenge claims they believe may be invalid. Indeed, from a competitive viewpoint, an insurer must pay only valid claims and must deny invalid claims to keep premiums to customers at a minimum. In a third-party case, both the insurer and the insured have a common interest in challenging a third-party’s claim. But in a first-party case, an insurer’s interest in challenging the claim directly conflicts with the insured’s interest in making the claim.
For another thing, insureds already have a remedy against their insurers by suit for breach of contract. Although this remedy does not ordinarily include consequential damages because of the rule of Hadley v. Baxendale, 9 Exch. 341, 354 (1854), or mental anguish damages, Stewart Title Guaranty Co. v. Aiello, 941 S.W.2d 68, 72 (Tex.1997), and thus does not always fully compensate an insured wrongfully deprived of benefits, Ashley, supra, at § 2.11, it is
just plain wrong to say “[wjithout a cause of action for breach of the duty of good faith and fair dealing ... insurers [can] ... arbitrarily den[y] coverage ... with no more penalty than interest on the amount owed.” [Union Bankers Ins. Co. v. Shelton, 889 S.W.2d 278, 283 (Tex.1994) (citing Arnold v. National County Mut. Fire Ins. Co., 725 S.W.2d 165, 167 (Tex.1987)) ]. This statement is no more true today than it was when it first was made. To the contrary, even if an insurer that denies a claim prevails at trial, it must pay the costs of its own defense. If the insured prevails, the insurer must pay not only prejudgment interest, but also the insured’s reasonable attorneys’ fees. Tex. Civ. Prac. & Rem. Code §§ 38.001-38.006 (Vernon 1986)... Further, an insurer may face statutory penalties for improper claims handling. See, e.g., Tex. Ins.Code Ann. arts. 21.21, 21.55 (Vernon 1981 & Supp.1994). Liabili*61ty for such additional sums, over and above policy benefits, may reasonably be expected to deter even “repeat-players” in litigation, like insurance carriers, from arbitrarily denying benefits under a policy.
Shelton, 889 S.W.2d at 286-287 (Cornyn, J., joined by Gonzalez and Hecht, JJ., concurring and dissenting).
Moreover, it is far from clear whether giving insureds a tort action does not create more problems than it solves. If the threat of liability is too great, insurers are discouraged from denying any claims, even those it suspects may be fraudulent.
This danger is doubly troubling in light of recent fears that insurance fraud is rampant. Although its exact magnitude is difficult to ascertain for obvious reasons, various estimates suggest that excess medical claims alone may add over $100 per year to the average cost of auto insurance for each customer, that about 10 percent of all auto insurance claims are fraudulent, that property/casualty insurance fraud may amount to about $20 billion annually, that health care fraud may be as much as $100 billion annually, and that workers’ compensation fraud may account for as much as 26 percent of all claims. It has even been suggested that insurance fraud is second only to illegal drug trafficking in the amount of illicit revenue that it generates for perpetrators. The potential joint returns to insurance companies and honest policyholders from reductions in fraud thus appear to be quite considerable across the board. If reasonable efforts to reduce fraud can become “bad faith” down the road, however, insurers may simply give up on them.
Alan O. Sykes, “Bad Faith” Breach of Contract by First-Party Insurers, 25 J. Legal Stud. 406, 434-435 (1996) (footnotes omitted).
In one of the few formal studies conducted, Massachusetts found that 32 percent of the total number of claims studied contained an element of fraud. An Insurance Research Council study estimates that 36 percent of all bodily injury liability claims appear to involve fraud or buildup. It is generally conceded by industry experts that 10 to 25 percent of all property-casualty claims are fraudulent.
Edward L. Schrenk & Jonathon B. Palmquist, Fraud and Its Effects on the Insurance Industry, 64 Def. CouNS. J. 23, 24 (1997) (footnotes omitted). If an insurer is to remain in business, compensation to bad-faith claimants, like policy benefits, must ultimately come from premiums paid by policyholders.
Still, courts have felt strong pressures to discourage unscrupulous insurers from taking advantage of their superior bargaining position to delay adjustment of claims and force insureds to accept less than they are entitled to. Ashley, supra, at § 2.14. Consequently, most jurisdictions have recognized a tort of bad-faith in first-party insurance cases. The first court of last resort to do so was the California Supreme Court in Gruenberg v. Aetna Insurance Co., 9 Cal.3d 566, 108 Cal.Rptr. 480, 510 P.2d 1032 (1973). Ashley, supra, at § 2.12; Roger C. Henderson, The Tort of Bad Faith in First-Party Insurance Transactions After Two Decades, 37 Ariz. L.Rev. 1153, 1153 (1995)[hereinafter Henderson I]; Roger C. Henderson, The Tort of Bad Faith in First-Party Insurance Transactions: Refining the Standard of Culpability and Reformulating the Remedies by Statute, 26 U. Mich. J.L. Ref. 1, 25-26 (1992)[hereinafter Henderson II]; Guy O. Kornblum, The Current State of Bad Faith and Punitive Damage Litigation in the U.S., 23 Tort & Ins. L.J. 812, 820 (1988); Richmond I, supra, at 104-105. Since then at least twenty-five other state supreme courts have recognized the tort in such cases. Chavers v. National Sec. Fire & Casualty Co., 405 So.2d 1 (Ala.1981); State Farm Fire & Casualty Co. v. Nicholson, 777 P.2d 1152 (Alaska 1989); Noble v. National Am. Life Ins. Co., 128 Ariz. 188, 624 P.2d 866 (1981); Aetna Casualty and Sur. Co. v. Broadway Arms Corp., 281 Ark. 128, 664 S.W.2d 463 (1984); Travelers Ins. Co. v. Servio, 706 P.2d 1258 (Colo.1985); Buckman v. People Express, Inc., 205 Conn. 166, 530 A.2d 596 (1987); The Best Place, Inc. v. Penn Am. Ins. Co., 82 Hawai’i 120, 920 P.2d 334 (1996); White v. Unigard Mut. Ins. Co., 112 Idaho 94, 730 P.2d 1014 (1986); Erie Ins. Co. v. Hickman, 622 N.E.2d 515 (Ind.1993); Dolan v. Aid Ins. Co., 431 N.W.2d 790 (Iowa 1988); *62Curry v. Fireman’s Fund Ins. Co., 784 S.W.2d 176 (Ky.1989); State Farm Fire & Casualty Co. v. Simpson, 477 So.2d 242 (Miss.1985); Lipinski v. Title Ins. Co., 202 Mont. 1, 655 P.2d 970 (1982) (common-law bad-faith tort actions against insurers abolished by Mont.Code Ann. § 38-18-242 (1993)); Braesch v. Union Ins. Co., 237 Neb. 44, 464 N.W.2d 769 (1991); United Fire Ins. Co. v. McClelland, 105 Nev. 504, 780 P.2d 193 (1989); State Farm Gen. Ins. Co. v. Clifton, 86 N.M. 757, 527 P.2d 798 (1974); Corwin Chrysler-Plymouth, Inc. v. Westchester Fire Ins. Co., 279 N.W.2d 638 (N.D.1979); Hoskins v. Aetna Life Ins. Co., 6 Ohio St.3d 272, 452 N.E.2d 1315 (1983); Christian v. American Home Assurance Co., 577 P.2d 899 (Okla.1978); Bibeault v. Hanover Ins. Co., 417 A.2d 313 (R.I.1980); Nichols v. State Farm Mut. Auto. Ins. Co., 279 S.C. 336, 306 S.E.2d 616 (1983); In re Certification of a Question of Law from the U.S. District Court, 399 N.W.2d 320 (S.D.1987); Arnold v. National County Mut. Fire Ins. Co., 725 S.W.2d 165 (Tex.1987); Anderson v. Continental Insurance Co., 85 Wis.2d 675, 271 N.W.2d 368 (1978); McCullough v. Golden Rule Ins. Co., 789 P.2d 855 (Wyo.1990). See Ashley, supra, at § 2.15; Barry R. Ostrager & Thomas R. Newman, Bad Faith and Wrongful Refusal to Settle: Liability in Excess of Policy Limits § 12.12(b), in Practising Law Institute, 13th Annual Insurance, Excess and Reinsurance Coverage Disputes 113, 167-172 (1996); Henderson I, supra, at 1153-1155; Henderson II, supra, at 26-28; Richmond I, supra, at 104-105.
Five other state supreme courts have allowed recovery of extra-contractual damages without recognizing the tort. Tackett v. State Farm Fire & Cas. Ins. Co., 653 A.2d 254 (Del.1995); Marquis v. Farm Family Mut. Ins. Co., 628 A.2d 644 (Me.1993); Lawton v. Great S.W. Fire Ins. Co., 118 N.H. 607, 392 A.2d 576 (1978); Beck v. Farmers Ins. Exch., 701 P.2d 795 (Utah 1985); Hayseeds, Inc. v. State Farm Fire & Casualty, 177 W.Va. 323, 352 S.E.2d 73 (1986). See Henderson I, supra, at 1154-1155; Henderson II, supra, at 28-29.
Four state courts of last resort have rejected a tort of bad faith in first-party insurance eases. Spencer v. Aetna Life & Casualty Ins. Co., 227 Kan. 914, 611 P.2d 149 (1980); Kewin v. Massachusetts Mut. Life Ins. Co., 409 Mich. 401, 295 N.W.2d 50 (1980); Haagenson v. National Farmers Union Property & Casualty Co., 277 N.W.2d 648 (Minn.1979); Halpin v. Prudential Ins. Co. of Am., 48 N.Y.2d 906, 425 N.Y.S.2d 48, 401 N.E.2d 171 (1979); Rocanova v. Equitable Life Assurance Soc’y, 83 N.Y.2d 603, 612 N.Y.S.2d 339, 634 N.E.2d 940 (1994); New York Univ. v. Continental Ins. Co., 87 N.Y.2d 308, 639 N.Y.S.2d 283, 662 N.E.2d 763 (1995). See also Ashley, supra, at § 2.15.
Despite fairly widespread adoption of the tort, two principal arguments continue to be made for rejecting altogether the tort of bad faith in first-party insurance cases.
First, most scholars have argued that contract law should provide all redress for an insurer’s wrongful denial of its insureds’ claims. See, e.g., Ashley, supra, at § 2.14; Mark Gergen, A Cautionary Tale About Contractual Good Faith in Texas, 72 Tex. L.Rev. 1235; William Powers, Jr., Border Wars, 72 Tex. L.Rev. 1209, 1229-1233 (1994); Sykes, supra, at 408-410. Reflecting on the adoption of the bad-faith tort in Texas, Professor Mark Gergen writes:
Opening the door to tort claims in contract, with their lure of emotional and exemplary damages, creates a crush of claims as plaintiffs and their lawyers attempt to cash in. Trial judges vary in how closely they police the door, and juries sometimes award extraordinary damages in what seem to be run-of-the-mill contract cases. While the Texas appellate courts have policed this particular door between contract and tort fairly closely (they limited the tort to the area of insurance, and only a few cases have upheld large exemplary damage awards when an insurer made what seemed a good faith mistake in denying a claim), the opinions are ill-reasoned because the ways in which the tort’s boundaries are defined defy reasoned analysis. The huge stakes in bad faith cases and its flimsy doctrinal garb make this a particularly volatile area of the law.
Gergen, supra, at 1236. The argument for a contract remedy rather than a tort aims *63mostly at the difference in available damages and the uncertainty in determining extraeon-tractual damages like mental anguish and punitive damages.
Second, the argument is made with increasing force that a bad-faith tort does more harm than good. Even scholars who accept the need for some remedy in tort are troubled by developments in bad-faith law:
The need for a tort remedy for insurers’ predatory or unscrupulous claims practices is well-recognized. But plaintiffs’ attorneys today strive to manipulate insurers into extracontractual and punitive exposure. Increasingly, plaintiffs groundlessly allege insurance companies’ bad faith in attempts to win the judicial lottery.... There now exists a nightmarish extracon-traetual insurance culture.
Richmond I, supra, at 76 (footnotes and text omitted). Professor Sykes writes:
Insurers may at times employ lamentable tactics to reduce their payments under first-party insurance policies, primarily by exploiting the delay inherent in the civil litigation process to induce a needy insured to settle for less than the amount that the contract promises. The prospect of tort remedies at the end of the litigation process can make such tactics unprofitable and thus serve a potentially valuable function. Yet we must be alert to the danger that the remedy may be worse than the problem — a fact that may explain the absence of greater market response to it. Nevertheless, one cannot rule out the possibility that the courts can step in constructively when an insurer knowingly refuses to make good on its clear obligations.
Unfortunately, however, the courts seem to find tortious conduct on the part of insurers who have bona fide disputes with their policyholders over the terms of the policy or over factual issues essential to the insured’s right to recover. The ability of the courts to identify opportunistic behavior in such cases is very much in doubt, and the distinct possibility arises that bad faith doctrine here does little to police misconduct while doing much to cause uneconomic increases in the premiums that policyholders must pay.
Sykes, supra, at 443.
Further complicating matters is the argument that bad faith cannot be a one-way street for liability only for insurers, especially in view of the pervasiveness of claims fraud. Insureds, too, it is said, should be liable for bad faith in appropriate circumstances. Douglas R. Richmond, The Two-Way Street of Insurance Good Faith: Under Construction, But Not Yet Open, 28 Loy. U. Chi. L.J. 95 (1996); Ellen Smith Pryor, Comparative Fault and Insurance Bad Faith, 72 Tex. L.Rev. 1505 (1994); Daniel S. Bopp, Tort and Contract in Bad Faith Cases: Is the Honeymoon Over?, 59 Def. Couns. J. 524 (1992). The difficulties in extending bad faith to insureds, however, press the question whether a bad-faith tort works in either direction. William Powers, Jr., What a Comparative Bad Faith Defense Tells Us About Bad Faith Insurance Litigation, 72 Tex. L.Rev. 1571 (1994).
While such arguments cannot be ignored, reversing course has its own problems. Bad faith has come to be accepted as a cause of action and is being litigated in a great many cases. The cause of action is currently recognized in about half the states; only a handful of jurisdictions have rejected it. And as virtually every legal commentator concedes, there should be some remedy for insurers’ abuse of the claims process. But if the tort is to exist, it must be defined.
B
“The term ‘bad faith’ ... is not self-defining, nor has it historically been a recognized independent basis of culpability in tort law. It has come to mean different things to different courts. Consequently, its use has caused definitional problems from the outset.” Henderson I, supra, at 1156 (footnotes omitted).
As in many jurisdictions, the tort of bad faith in Texas has two elements: “that the insurer had no reasonable basis for denying or delaying payment of the claim, and that it knew or should have known that fact.” Transportation Ins. Co. v. Moriel, 879 S.W.2d 10, 18 (Tex.1994); accord Republic *64Ins. Co. v. Stoker, 903 S.W.2d 338, 340 (Tex.1995); Shelton, 889 S.W.2d at 283; Aranda v. Insurance Co. of N. Am., 748 S.W.2d 210, 213 (Tex.1988); see Arnold, 725 S.W.2d at 167; Ashley, supra, at § 5.02. Thus, the touchstone of bad-faith liability — or more precisely, a breach of the duty of good faith and fair dealing — is unreasonableness in processing insurance claims. Richmond I, supra, at 109. The measure of reasonableness is not clear. It is the courts’ duty to provide clarity because the existence and scope of a legal duty are questions of law. Otis Engineering Corp. v. Clark, 668 S.W.2d 307, 309 (Tex.1983). The courts have failed to define the scope of the duty.
Unreasonableness in the bad-faith context could be construed to mean doing what an ordinary person, or perhaps an ordinary insurer, would not do. More strictly, it might mean doing what no ordinary person or insurer would do. All four of these possibilities are essentially negligence standards. If reference were to an ordinary insurer, then expert testimony would be necessary to prove liability. Timothy J. Muldowney & Robert A. Zupkus, Bad Faith Claims: The Role of the Expert, 64 Def. Couns. J. 226 (1997); James C. Peterson, Selecting and Using Expert Witnesses in Bad Faith Cases, 19 Am. J. Trial Advoo. 543 (1996). Bad faith would be insurance malpractice. If reference were only to an ordinary person, expert testimony would not be necessary. Bad faith would be simple negligence. There are at least four reasons why the unreasonable conduct that constitutes bad faith cannot be mere negligence or malpractice.
First, to equate bad faith with negligence or malpractice would mean that an insurer would be hable for failing to pay a claim that an insurer using ordinary care would have paid. An insurer using ordinary care would pay the claims it was legally obligated to pay. Completing the syllogism, bad faith would mean failing to pay the claims an insurer is legally obligated to pay. But such failure is breach of contract, and we have reiterated that bad faith cannot be equated to breach of contract. Moriel, 879 S.W.2d at 17-18; National Union Fire Ins. Co. v. Dominguez, 873 S.W.2d 373, 376-377 (Tex.1994). Hence, there must be something more to bad faith than failing to do what an ordinarily prudent insurer would do. As we said in Moriel, “[t]he threshold of bad faith is reached when a breach of contract is accompanied by an independent tort.” Id. The essence of that independent tort is the unscrupulous taking advantage of an insured’s disadvantageous position relative to the insurer.
Second, the realities of the insurance market and industry make it virtually impossible to ascertain and apply a standard of care for handling claims.
[Liability ought not depend on a finding that an insurer mishandled a claim, as it does under the Texas common law and statutory actions that make liability dependent on a finding of bad faith, because the standard of care we should demand of insurers in processing doubtful claims is unclear. Plainly, there is some limit on what insurers ought to spend to investigate claims. For instance, it would be folly to require an insurer to spend thousands of dollars to investigate a claim worth only hundreds of dollars when the expenditure would provide information that would only slightly increase the accuracy of the decision to grant or deny the claim. In determining proper claims procedure, we must balance the value of the claim, the value of accurate claim determination, the cost of additional precaution, and the improvement in the accuracy of claim determination from additional precaution. I do not know how to value accurate claim determination because accuracy has noneconomic value in terms of respect for individual rights and it is difficult to assess economic value in the effect on insurance holders in taking precautions and purchasing insurance. Also, even if we knew what the optimal balance of these variables was in theory, we could not implement that standard in practice because courts and juries would be unable to measure the relevant variables.
Gergen, supra, at 1257-1258.
Third, even if a standard of ordinary care could be ascertained, the duty of good faith and fair dealing may hold insurers to a higher responsibility than is common in the in*65dustry. The fact that insurers ordinarily processed claims without regard to the facts — if that were what they did — would not mean that they acted in good faith. In creating the duty of good faith and fair dealing, the law has not simply obligated insurers to abide by industry standards in processing claims; rather, it has undertaken to set those standards.
Fourth, the purpose of the duty of good faith and fair dealing is to discourage “unscrupulous insurers [from taking] advantage of their insureds' misfortunes in bargaining for settlement or resolution of claims.” Arnold, 725 S.W.2d at 167; Allstate Insurance Co. v. Watson, 876 S.W.2d 145, 148 (Tex.1994); Natividad v. Alexsis, Inc., 875 S.W.2d 695, 697-698 (Tex.1994); Lyons v. Millers Cos. Ins. Co., 866 S.W.2d 597, 599-600 (Tex.1998). The duty is also to prevent insurers from acting arbitrarily. Arnold, 725 S.W.2d at 167. “Unscrupulous”, “arbitrary”, and “taking advantage” are not synonyms for negligence. They require, not merely misfeasance, but malfeasance.
The distinction between the unreasonableness standard for bad faith and the unreasonableness standard in other liability contexts is significant, and the confusion of the two concepts accounts for much of the present difficulty with the bad-faith tort. Defining bad faith as a failure to use ordinary care is simply unworkable.
The judicially created rules of liability in bad faith eases [based on ordinary care] presupposes a model of insurance company behavior that does not and cannot exist. The courts suppose that an insurance company can respond to any claim by immediately appointing a well-educated, well-trained claims adjuster to investigate the claim in minute detail, come to a reasoned and balanced conclusion about the facts, reach a legally correct interpretation of the policy, and promptly pay the claim or explain to the insured why the insurer denied the claim. In fact, the economics of the claims settlement process dictate that insurance companies delegate a major portion of claims settlement work to low-level claims agents, sometimes inexperienced or poorly trained, whose workloads preclude their giving the attention to all claims that the judicial model expects.
Ashley, supra, at § 10.28; accord, Richmond I, supra, at 140.
But if bad faith is not negligence, what is it? The failure to define clear liability standards for a bad-faith tort has caused many problems. One is that insurers cannot anticipate what circumstances will risk liability and avoid them.
[W]here the central element of bad faith has no clearly defined structure, then virtually anything becomes relevant and specific conduct can be judged only in retrospect. Thus, an insurer has little ability to evaluate the potential consequences of its elaims-handling decisions, other than those which acknowledge full coverage irrespective of the terms of its policy.
James C. Nielsen, Advice of Counsel in Insurance Bad Faith Litigation: A Substantive Framework for Pleading, Discovery and Proof, 25 TORT & Ins. L.J. 533, 536 (1990). It is fundamentally unfair for the law to impose liability without defining what conduct is culpable precisely enough for it to be avoided.
Another problem with not defining bad faith is that it encourages the allegation in every case.
Unfortunately, insurance bad faith litigation has gotten out of hand. When courts initially recognized bad faith as an independent tort they anticipated that such claims would be rare. In fact, they should be. By and large insurers process a vast amount of claims in an ethical and responsible manner. But now, almost every first-party action for an insurer’s breach of contract includes a bad faith count, and liability insurers are deliberately “set up” for bad faith claims.
Richmond I, supra, at 140 (footnotes omitted). Commentators refer to bad-faith actions as “the judicial equivalent of the Wheel of Fortune”, id., and a “lottery”, Douglas G. Houser, Good Faith as a Matter of Law: The Insurance Company’s Right to Be Wrong, 27 Tort & Ins. L.J. 665, 666 (1992). There is always a chance it will pay off. But inclusion of a bad-faith allegation in every first-party insurance case distorts not only *66the litigation system but the insurance industry itself.
I am inclined to think that a bad-faith count is included in most claims against insurers in jurisdictions that recognize the tort. We must assume that the count adds a premium to whatever value the claim has, and this suggests that the overall, system-wide cost of recognizing the tort of bad faith may be large, even if insurers are now able to better identify and redistribute these costs.
Robert H. Jerry, II, The Wrong Side of the Mountain: A Comment on Bad Faith’s Unnatural History, 72 Tex. L.Rev. 1317, 1343 (1994). Bad-faith allegations not only make cases more expensive to settle, “[t]he addition of a claim of uncertain value in a dispute makes settlement less likely and thus increases litigation costs.” Gergen, supra, at 1252-1253. See Houser, supra, at 665.
Still another problem with an undefined bad-faith tort is the effect on policyholders and the public who never have a claim.
There is no doubt that the tort of bad faith has had and will continue to have a terrific impact on first-party insurers. In practically every lawsuit that is filed against an insurer providing first-party benefits there is a count seeking consequential and punitive damages, in addition to policy benefits. This development has significant implications for insurers and how they do business.
Henderson I, supra, at 1182. Bad-faith liability “enriches a few at the expense of many with significant transaction costs.” Gergen, supra, at 1250. Without principles and limits, a bad-faith tort does more harm than good.
In the context of insurance claims practices, it is clear that some measures were needed to redress the legitimate complaints of insureds, but now it appears that the balance may have been tipped too far in their favor by unduly exposing insurers to extra-contractual damages. The judicial filling of the vacuum left by the legislative and administrative processes was justified when it was made; real abuses by insurers need to be identified and corrected. Nevertheless, the new tort remedy, although necessary in some form, now shows signs of being too oppressive on an industry whose financial vitality and efficiency are essential to social well-being. Multimillion dollar awards for wrongfully denying claims not only are unnecessary to correct the situation, but such awards, which often have a windfall nature, may raise the cost of insurance for the vast numbers of insureds who are not mistreated and may do great harm to the risk-transfer-and-distribution mechanism in our society by making insurance so expensive that it can no longer be purchased like a household commodity. There is a point at which potential insureds will either elect reduced coverage or forgo purchases or other activities because of insurance costs. This negative impact certainly could extend to and affect the standard of living for individuals if too much of their income must be spent on premiums that spread the cost of awards for extra-contractual damages and the related expenses of defending against such claims, in addition, to covering the primary risks insured against.
Henderson II, supra, at 31-32 (footnote omitted).
Improvement in processing first-party insurance claims should not be more than an incidental burden on the provision of insurance. If insurers know what conduct to avoid, they -can minimize the risks of bad-faith liability. However, if insurers cannot know what constitutes bad faith until a verdict is returned, the costs are unavoidable.
Uncertainty is not bad faith’s only vice. “A mature body of law is more than just predictable, of course. It embodies a stable, coherent, and defensible logic, and it grows incrementally as judges who understand and accept the logic carefully apply it to new cases.” Ellen Smith Pryor & Charles Silver, Introduction to the Symposium on the Law of Bad Faith in Contract and Insurance, 72 Tex. L.Rev. 1203 (1994). This does not describe the bad-faith tort in first-party insurance eases. Rather:
Although the courts responded to the need for relief from the unfair claims practices of insurers, the common-law remedy has foundered because of the judicial ina*67bility to strike an appropriate balance between the parties to an insurance contract. A substantial number of jurisdictions have purported to adopt the tort of bad faith, but there has been altogether too little attention given to defining the standard of culpability.
Henderson II, supra, at 66. See Bopp, supra, at 524 (“[B]ad-faith insurance law has remained in perpetual transition.”)
The central idea of bad-faith liability— acting without any reasonable basis — requires more definition that this Court has yet given it.
C
Like many courts, our principal reason for adopting the bad-faith tort was to discourage “unscrupulous insurers [from taking] advantage of their insureds’ misfortunes in bargaining for settlement or resolution of claims.” Arnold, 725 S.W.2d at 167; Watson, 876 S.W.2d at 148; Natividad, 875 S.W.2d at 697-698; Lyons, 866 S.W.2d at 599-600.
It is not unscrupulous for an insurer to deny or delay payment of a claim about which there is a bona fide dispute. Moriel, 879 S.W.2d at 17. We have never intended that an insurer should be liable for tort damages simply for denying a claim in error, even if the insurer was negligent. To the contrary, we promised at the tort’s inception that “carriers will maintain the right to deny invalid or questionable claims and will not be subject to liability for an erroneous denial of a claim.” Aranda, 748 S.W.2d at 213. We have reiterated the promise in several cases.
In other words, if the insurer has denied what is later determined to be a valid claim under the contract of insurance, the insurer must respond in actual damages up to the policy limits. But as long as the insurer has a reasonable basis to deny or delay payment of the claim, even if that basis is eventually determined by the factfinder to be erroneous, the insurer is not liable for the tort of bad faith.
Lyons, 866 S.W.2d at 600.
Evidence that merely shows a bona fide dispute about the insurer’s liability on the contract does not rise to the level of bad faith.... Nor is bad faith established if the evidence shows the insurer was merely incorrect about the factual basis for its denial of the claim, or about the proper construction of the policy.... A simple disagreement among experts about whether the cause of the loss is one covered by the policy will not support a judgment for bad faith.
Moriel, 879 S.W.2d at 17-18.
“[This] bona fide dispute rule” is nothing more than a shorthand notation for the observation that the parties to an insurance contract will sometimes have a good faith disagreement about coverage. Under such circumstances, the parties may require a court to interpret policy language for them, or a jury to resolve factual disputes. Simply because the parties go to court does not raise an issue of the insurer’s bad faith.
Id. at 18 n. 8. Accord, Stoker, 903 S.W.2d at 340-41; Shelton, 889 S.W.2d at 287 (Cornyn, J., joined by Gonzalez and Hecht, JJ., concurring and dissenting) (“Carriers must maintain the right to deny invalid or questionable claims and not be subject to bad-faith liability for an erroneous denial of a claim.”).
In theory, what protects an insurer from bad-faith liability for simply denying a claim erroneously is the first element of the tort—no reasonable basis. Stoker, 903 S.W.2d at 340; Texas Farmers Ins. Co. v. Soriano, 881 S.W.2d 312, 319 (Cornyn, J., joined by Hecht, J., concurring). This is true in most other jurisdictions that have recognized a bad-faith tort. An early first-party bad-faith case, Anderson v. Continental Insurance Co., 85 Wis.2d 675, 271 N.W.2d 368, 376-377 (1978), stated what seems intuitive, that insurers have a reasonable basis for challenging claims that are “fairly debatable” and thus cannot be liable for bad faith for doing so. Anderson, 271 N.W.2d at 376-377. Bad faith does not extend to all errors in judgment. An insurer has the “right to be wrong”. Richmond I, supra, at 109; William T. Barker & Paul E.B. Glad, Use of Summary Judgment in Defense of Bad Faith Actions In*68volving First-Party Insurance, 30 Tort & Ins. L.J. 49, 51 (1994); Houser, supra, at 666.
The supreme courts in at least seventeen states that have recognized either a tort of bad faith or liability for extra-contractual damages in first-party insurance settings have expressly endorsed the “fairly debatable” exception to liability. Gilbert v. Congress Life Ins. Co., 646 So.2d 592, 593 (Ala.1994); National Sav. Life Ins. Co. v. Dutton, 419 So.2d 1357 (Ala.1982); National Security Fire & Cas. Co. v. Bowen, 417 So.2d 179, 183 (Ala.1982); Rawlings v. Apodaca, 151 Ariz. 149, 726 P.2d 565 (1986); Noble, 624 P.2d at 868; Savio, 706 P.2d at 1275; Garnett v. Transamerica Ins. Servs., 118 Idaho 769, 800 P.2d 656, 666 (1990); White, 730 P.2d at 1020; Morgan v. American Family Mut. Ins. Co., 534 N.W.2d 92, 96 (Iowa 1995); Dolan, 431 N.W.2d at 794; Curry, 784 S.W.2d 176; Andrew Jackson Life Ins. Co. v. Williams, 566 So.2d 1172, 1184-1185 (Miss.1990); Tynes v. Bankers Life Co., 224 Mont. 350, 730 P.2d 1115, 1124 (1986); Braesch, 464 N.W.2d at 777-778; Pickett v. Lloyd’s, 131 N.J. 457, 621 A.2d 445, 453 (1993); Tokles & Son, Inc. v. Midwestern Indem. Co., 65 Ohio St.3d 621, 605 N.E.2d 936, 943 (1992); Rumford Prop. & Liab. Ins. Co. v. Carbone, 590 A.2d 398, 400 (R.I.1991); Bibeault v. Hanover Ins. Co., 417 A.2d 313, 319 (R.I.1980); In re Certification of a Question of Law from the U.S. District Court, 399 N.W.2d at 324; Billings v. Union Bankers Ins. Co., 918 P.2d 461, 464 (Utah 1996); Beck v. Farmers Ins. Exch., 701 P.2d 795 (Utah 1985); Bushey v. Allstate Ins. Co., 164 Vt. 399, 670 A.2d 807, 809 (1995); Warmka v. Hartland Cicero Mut. Ins. Co., 136 Wis.2d 31, 400 N.W.2d 923, 925 (1987); Anderson, 271 N.W.2d at 377; State Farm Mut. Auto. Ins. Co. v. Shrader, 882 P.2d 813, 825 (Wyo.1994); McCullough, 789 P.2d at 860-861.
A number of federal appellate courts applying state law have also recognized this principle. State Farm Fire & Casualty Co. v. Balmer, 891 F.2d 874 (11th Cir.1990) (Alabama law); Haynes v. Allstate Ins. Co., 749 F.2d 1474 (11th Cir.1985) (Alabama law); Dempsey v. Auto Owners Ins. Co., 717 F.2d 556 (11th Cir.1983) (Alabama law); Lange v. Penn Mut. Life Ins. Co., 843 F.2d 1175 (9th Cir.1988) (Arizona law); Chadima v. National Fidelity Life Ins. Co., 55 F.3d 345 (8th Cir.1995) (Iowa law); Wolf v. Prudential Ins. Co. of Am., 50 F.3d 793 (10th Cir.1995) (Oklahoma law); Bilden v. United Equitable Ins. Co., 921 F.2d 822 (8th Cir.1990) (North Dakota law); Pace v. Insurance Co. of N. Am., 838 F.2d 572 (1st Cir.1988) (Rhode Island law); Case v. Toshiba Am. Info. Sys., Inc., 7 F.3d 771 (8th Cir.1993) (South Dakota law); Kirchojf v. American Casualty Co., 997 F.2d 401 (8th Cir.1993) (South Dakota law); American Casualty Co. v. B. Cianciolo, Inc., 987 F.2d 1302 (7th Cir.1993) (Wisconsin law); Duir v. John Alden Life Ins. Co., 754 F.2d 245 (7th Cir.1985) (Wisconsin law).
No federal circuit and only one state supreme court has rejected the idea that insurers should not be liable for denying or delaying payment of fairly debatable claims. State Farm Mut. Auto. Ins. Co. v. Laforet, 658 So.2d 55, 62-63 (Fla.1995). But see William T. Barker, What’s New and Important in Bad Faith, 6 No. 1 COVERAGE 40 (1996) (arguing that the Montana Supreme Court has also rejected the “fairly debatable” standard in Palmer v. Farmers Ins. Exch., 261 Mont. 91, 861 P.2d 895 (1993); Dean v. Austin Mut. Ins. Co., 263 Mont. 386, 869 P.2d 256 (1994); and De Bruycker v. Guaranty Nat’l Ins. Co., 266 Mont. 294, 880 P.2d 819 (1994)).
To determine whether a claim is fairly debatable, the Alabama Supreme Court developed what has come to be called the “directed verdict rule”:
In the normal ease in order for plaintiff to make out a prima facie case of bad faith refusal to pay an insurance claim, the proof offered must show that the plaintiff is entitled to a directed verdict on the contract claim and, thus, entitled to recover on the contract claim as a matter of law. Ordinarily, if the evidence produced by either side creates a fact issue with regard to the validity of the claim and, thus, the legitimacy of the denial thereof, the tort claim must fail and should not be submitted to the jury.
Dutton, 419 So.2d at 1362. As explained by one commentator:
*69The logic of the Dutton rule is simple. An insurer is entitled to dispute claims so long as it has a reasonable basis. If reasonable minds could not differ on the coverage-determining facts, a verdict should be directed or summary judgment rendered on coverage. If that cannot be done, it ordinarily must follow that the insurer had reasonable grounds to dispute the facts, precluding any possibility of bad faith.
Barker, supra, at 56. See also Richmond I, supra, at 122; Stephen D. Heninger & Nicholas W. Woodfield, A Practitioner’s Guide to Alabama’s Tort of Bad Faith, 57 Ala. Law. 277 (1996). The directed-verdict rule in Dut-ton, with some specific exceptions, id. at 68-72, is followed in a number of jurisdictions. E.g., Curry, 784 S.W.2d 176; Andrew Jackson Life Ins. Co., 566 So.2d at 1184-1185; Carbone, 590 A.2d at 400. See also Ashley, supra, at § 5.03.50; Houser, supra, at 669.
D
While this Court has not previously used the phrase, “fairly debatable”, I believe it accurately characterizes those claims decisions for which insurers should not incur bad-faith liability. The Court has repeatedly reaffirmed that such decisions exist. The phrase, used in most states which have adopted the tort, adds definition to its “no reasonable basis” element.
I would adopt the “fairly debatable” standard for determining bad faith but for the recent intervention of the Legislature in this area of the law. In 1995, the Legislature amended Article 21.21, Section 4, of the Insurance Code to define unfair insurance settlement practices. The amendment provides in part:
See. 4. The following are hereby defined as unfair methods of competition and unfair and deceptive acts or practices in the business of insurance:
$ ‡ ‡
(10) Unfair Settlement Practices, (a) Engaging in any of the following unfair settlement practices with respect to a claim by an insured or beneficiary:
ij: j*« ifc s}:
(ii) failing to attempt in good faith to effectuate a prompt, fair, and equitable settlement of a claim with respect to which the insurer’s liability has become reasonably clear....
Tex. Ins.Code art. 21.21, § 4(10)(a)(ii). Section 16 of Article 21.21 gives a private right of action for violation of Section 4.
Section 4(10) is similar to Article 21.21-2, Section 2, which also prohibits unfair claim settlement practices, including “[n]ot attempting in good faith to effectuate prompt, fair, and equitable settlements of claims submitted in which liability has become reasonably clear”. Id. art. 21.21-2, § 2(b)(4). But there is no private cause of action under Article 21.21-2. American Physicians Ins. Exch. v. Garcia, 876 S.W.2d 842, 847 n. 11 (Tex.1994); Watson, 876 S.W.2d at 148. In Vail v. Texas Farm Bureau Mutual Insurance Company, 754 S.W.2d 129 (Tex.1988), the Court held that Article 21.21-2 could be used as a standard of liability in other causes of action. Enactment of Section 4(10) marks the first time the Legislature has created a private cause of action specifically for statutorily-defined unfair insurance claims settlement practices.
One of the prerequisites for liability under Section 4(10)(a)(ii) is that “the insurer’s liability has become reasonably clear”. This standard, like the “fairly debatable” standard, defines when denial of a claim is no longer merely erroneous but in bad faith. The same idea underlies both standards. Liability for payment of a claim is reasonably clear when it is no longer fairly debatable. As long as liability for a claim is fairly debatable, and hence not reasonably clear, an insurer may deny the claim without acting in bad faith.
Section 4(10) applies only to causes of action that either accrued after September 1, 1995, or accrued before that date but were not filed until after it. Act of May 19, 1995, 74th Leg., R.S., ch. 414, § 20,1995 Tex. Gen. Laws 2988, 3004. Therefore, it does not apply to this case. Nevertheless, I would join Justice Spectok in adopting the Legislature’s “reasonably clear” standard as the standard for the common-law bad-faith tort. *70To adopt the “fairly debatable” standard would be needlessly confusing, especially since the two standards are essentially identical. Moreover, inasmuch as the purpose of Article 21.21 is “to regulate trade practices in the business of insurance by defining, or providing for the determination of, all such practices in this state which constitute unfair methods of competition or unfair or deceptive acts or practices and by prohibiting the trade practices so defined or determined”, the statute is the exclusive remedy for bad faith after its effective date (emphasis added).
E
Although the Court has treated the “no reasonable basis” element of the bad-faith tort as a question of fact, not until Nicolau, which the Court also decides today, and Simmons has the Court been asked directly to consider whether the issue should be one of law for the court. I believe the issue should be one of law for several reasons.
First, the determinative standard is whether liability for a claim has become “reasonably clear”. In other contexts, whether something is reasonably clear is a legal issue, not a factual one. Whether a contract is unambiguous, and therefore reasonably clear, is a legal issue. R & P Enterprises v. La Guarta, Gavrel & Kirk, Inc., 596 S.W.2d 517, 518-519 (Tex.1980). A trust is created when the intent to do so appears reasonably clear, a legal issue. Dulin v. Moore, 96 Tex. 135, ■ 70 S.W. 742, 743 (1902); Perfect Union Lodge v. Interfirst Bank, 748 S.W.2d 218, 220 (Tex.1988). Whether an administrative regulation is sufficiently specific to satisfy constitutional due process concerns depends on whether it is reasonably clear, a legal issue. Jordan v. State Bd. of Ins., 160 Tex. 506, 334 S.W.2d 278, 280 (1960). The same is true of penal statutes, Fogo v. State, 830 S.W.2d 592, 595 (Tex.Crim.App.1992) (citing Kolender v. Lawson, 461 U.S. 352, 103 S.Ct. 1855, 75 L.Ed.2d 903 (1983)), and criminal indictments, Murk v. State, 775 S.W.2d 415, 418 (Tex.App. — Dallas 1989).
Second, whether liability on a claim is fairly debatable it treated as a legal issue in several states that apply that standard. Morgan, 534 N.W.2d at 96; Dolan, 431 N.W.2d at 794; Curry, 784 S.W.2d 176; Andrew Jackson Life Ins. Co., 566 So.2d at 1184-1185; Carbone, 590 A.2d at 400; Billings, 918 P.2d at 464 (contract cause of action); Beck, 701 P.2d 795. Although a few courts appear to treat “fairly debatable” as a fact question, it is not clear whether they would hold that the issue is itself a question of fact or only that it may involve questions of fact. See Tynes, 730 P.2d at 1124; Pickett, 621 A.2d at 453; Tokles & Son, 605 N.E.2d at 943; In re Certification of a Question of Law from the U.S. District Court, 399 N.W.2d at 324; Warmka, 400 N.W.2d at 925; Anderson, 271 N.W.2d at 377; Shrader, 882 P.2d at 825; McCullough, 789 P.2d at 860-861. Other states that recognize bad-faith liability have not addressed the issue.
Third, the issue, as I have noted, is not whether an insurer acted reasonably in the sense of without negligence, that is, with the care of an ordinarily prudent insurer in the same situation. This would be a fact issue which juries can readily decide. But proof of bad faith does not require a showing that the insurer departed from the standard of conduct common to the industry, although such evidence may be relevant. Rather, the issue is whether liability for the claim was reasonably clear. It may be necessary for the finder of fact to resolve factual disputes first about what information was known to the insurer, but once they are resolved, whether liability for the claim became reasonably clear is essentially a legal question.
Fourth, treating the issue as one of law allows the courts to begin to categorize the conduct which risks bad-faith liability and thus develop the practical parameters of the tort. For example, we stated in Moriel that “[a] simple disagreement among experts about whether the cause of the loss is one covered by the policy will not support a judgment for bad faith.” Moriel, 879 S.W.2d at 18. A case of this type is not bad faith, even though the insured’s experts testify that the insurer’s position was unreasonable. By the same token, the mere fact that an insurer can adduce expert testimony to support its position may not insure it from liability if the testimony is pretextual. By defining specific *71types of conduct or situations which do or do not give rise to bad-faith liability, the courts can begin to give the certainty and predictability to the tort that it must have and has not yet acquired.
Finally, determining what is “reasonably clear” amounts to defining an insurer’s legal duty. The definition of a legal duty is an issue for the court involving questions of legal policy. E.g., Graff v. Beard, 858 S.W.2d 918, 919-920 (Tex.1998). Other policy-laden matters are also legal issues for the court rather than fact issues for a jury. E.g., Dresser Indus., Inc. v. Page Petroleum, Inc., 853 S.W.2d 505, 509 (Tex.1993) (whether language is conspicuous); Donwerth v. Preston II Chrysler-Dodge, Inc., 775 S.W.2d 634, 637 (Tex.1989) (whether a cause of action is groundless and brought in bad faith under the Deceptive Trade Practices Act); Musser v. Smith Protective Servs., Inc., 723 S.W.2d 653, 654-655 (Tex.1987) (whether language is defamatory); Akin v. Dahl, 661 S.W.2d 917, 920-921 (Tex.1983), cert. denied, 466 U.S. 938, 104 S.Ct. 1911, 80 L.Ed.2d 460 (1984) (whether there was probable cause for prosecution).
As in other instances, the “reasonably clear” issue may turn on factual disputes that must first be resolved by the finder of fact. For example, whether an insurer has a reasonable basis for handling a claim may depend on whether it actually did the things it claims to have done, matters which may be factually disputed. Such disputes must, of course, be resolved by the finder of fact. Once resolved, however, the issue of whether the insurer had a reasonable basis for its actions is a legal one. The second element of the bad-faith tort — the requisite level of intent by the insurer — is, of course, a fact question.
II
The second element of bad faith is that an insurer “knew or should have known” that it had no reasonable basis for denying or delaying payment of a claim. Moriel, 879 S.W.2d at 18; accord Stoker, 903 S.W.2d at 340; Shelton, 889 S.W.2d at 283; Aranda, 748 S.W.2d at 213; see Arnold, 725 S.W.2d at 167. The Court has not been precise in stating what mental state an insurer must have for bad faith.
The Restatement differentiates between intentional, reckless and negligent conduct. “The word ‘intent’ ... denote[s] that the actor desires to cause consequences of his act, or that he believes that the consequences are substantially certain to result from it.” Restatement (Second) op ToRts § 8A (1965). An actor is reckless
if he does an act or intentionally fails to do an act which it is his duty to the other to do, knowing or having reason to know of facts which would lead a reasonable man to realize, not only that his conduct creates an unreasonable risk of physical harm to another, but also that such risk is substantially greater than that which is necessary to make his conduct negligent.
Id. at § 500. The contrast between intent and recklessness has been described as follows:
[The Restatement] definition [of recklessness] encompasses two different states of mind. First, when a person acts, or fails to act, with awareness of the high degree of risk, the conduct is reckless. Second, when a person acts with awareness of the facts which give rise to the risk, but is not aware of the high degree of risk involved, that conduct still is considered reckless if a reasonable person in his or her place would be aware of such risk. According to the Restatement, the essential difference between intentional and reckless conduct lies in the level of the actor’s knowledge of the degree of certainty that consequences will flow from the actor’s conduct. Both intentional and reckless conduct involve intentional acts, but the conduct is classified as “intentional” only when the actor knows or knows with substantial certainty that harm will ensue from the act. If the actor knows or, from facts which she possesses, has reason to know that there is only a very strong probability, and not a substantial certainty, that harm may result from the act, the conduct is classified as “reckless.”
Henderson II, supra, at 44 (footnotes omitted).
*72“Reason to know5’, a recklessness standard, is not equivalent to “should know”, a negligence standard.
The words “reason to know” ... denote the fact that the actor has information from which a person of reasonable intelligence or of the superior intelligence of the actor would infer that the fact in question exists, or that such person would govern his conduct upon the assumption that such fact exists.
The words “should know” ... denote the fact that a person of reasonable prudence and intelligence or of the superior intelligence of the actor would ascertain the fact in question in the performance of his duty to another, or would govern his conduct upon the assumption that such fact exists.
Restatement (Seoond) of ToRts § 12 (1965). The Restatement contrasts recklessness and negligence as follows:
Reckless misconduct differs from negligence in several important particulars. It differs from that form of negligence which consists in mere inadvertence, incompetence, unskillfulness, or a failure to take precautions to enable the actor adequately to cope with a possible or probable future emergency, in that reckless misconduct requires a conscious choice of a course of action, either with knowledge of the serious danger to others involved in it or with knowledge of facts which would disclose this danger to any reasonable man. It differs not only from the above-mentioned form of negligence, but also from that negligence which consists in intentionally doing an act with knowledge that it contains a risk of harm to others, in that the actor to be reckless must recognize that his conduct involves a risk substantially greater in amount than that which is necessary to make his conduct negligent. The difference between reckless misconduct and conduct involving only such a quantum of risk as is necessary to make it negligent is a difference in the degree of the risk, but this difference of degree is so marked as to amount substantially to a difference in kind.
Id. at § 500 cmt. g.
These distinctions are useful in defining bad faith. Bad faith clearly encompasses an insurer’s denying or delaying payment of a claim when the insurer knows, at least to a substantial certainty, that it has no reasonable basis for doing so. An insurer should also be hable when it is aware that there is a high degree of risk that it has no reasonable basis for denying or delaying payment of its insured’s claim. This is the first prong of recklessness defined in the Restatement.
[I]t would be particularly offensive for an insurance company to appreciate the fact that the claim should be paid, but to refuse to do so on the slim hope that it might succeed in defending a lawsuit brought to collect the proceeds. Such conduct is precisely the kind of conduct that the tort of bad faith should reach because it serves no useful purpose and is destructive to the insured. This behavior presents the strongest possible case for recovery of extra-contractual damages. Failure to recognize a cause of action for such arbitrary conduct not only would frustrate the process of transferring and spreading risks of loss in our society, but it also would permit insurers actually to profit from their wrongful conduct by allowing them to earn income on funds that rightfully belong to their insureds.
Henderson II, supra, at 50 (footnote omitted).
The first prong of reckless, like the definition of intent, is subjective — it depends on what an actor knows or is aware of. Awareness of likelihoods is less than knowledge to at least a substantial certainty, but it is nevertheless an actual consciousness. The second prong of recklessness does not share this feature with intent. It is objective: the actor is not actually aware of the high degree of risk involved in his actions, but he knows or has reason to know facts which would cause a reasonable person in his position to be aware of the high degree of risk.
The required mental state for bad-faith liability should include the second prong of recklessness. Insurers should be experts in processing claims. Insureds ordinarily have no such knowledge. In balancing the rights of insurers and insureds, it is reasonable to require insurers to take responsibility for *73denying or delaying payment of claims when they have reason to know there is a high degree of risk that there is no reasonable basis for doing so.
Bad-faith liability should not, however, extend to negligence — an insurer’s failure to pay a claim as an ordinarily prudent insurer would. For one thing, it is not clear what the standard of ordinary care would be. As Professor Gergen argues, the decision to pay claims usually involves a number of variables which are difficult to evaluate. Payment of even a small claim is often important to an insured, but thei’e are practical limits to the time and resources an insurer can expend in investigating a claim. For another thing, even if a negligence standard could practicably be determined, insurers should be permitted some margin of error before suffering extra-contractual damages. “Insureds do make inaccurate, groundless, and even fi’aud-ulent claims and insurers need some margin of error within which to operate, but this margin should not extend into the area of conduct described as reckless under the Restatement of Torts.” Henderson II, supra, at 54.
I would hold that “knew or should have known” in the second element of bad faith requires intentional or reckless conduct as defined in the Restatement. This appears to be the rule in a majority of jurisdictions. Most states that have addressed the issue— at least ten — appear to require intentional or reckless conduct for liability. Gulf Atl. Life Ins. Co. v. Barnes, 405 So.2d 916, 924 (Ala.1981); Nicholson, 777 P.2d at 1154 n. 3; Deese v. State Farm, Mut. Auto. Ins. Co., 172 Ariz. 504, 838 P.2d 1265, 1268 (1992); Savio, 706 P.2d at 1274 (used by legislature when bad faith codified); Curry, 784 S.W.2d at 178; Braesch, 464 N.W.2d at 778; Bibeault, 417 A.2d at 319; In re Certification of Question of Law in the U.S. District Court, 399 N.W.2d at 324; Fehring v. Republic Ins. Co., 118 Wis.2d 299, 347 N.W.2d 595, 600-601 (1984); McCullough, 789 P.2d at 860. Two appear to impose bad-faith liability for gross negligence. Aetna Casualty & Sur. Co. v. Day, 487 So.2d 830, 832 (Miss.1986); Jessen v. National Excess Ins. Co., 108 N.M. 625, 776 P.2d 1244, 1247 (1989). Three others impose liability for negligence. Kiner v. Reliance Ins. Co., 463 N.W.2d 9, 12 (Iowa 1990); Zoppo v. Homestead Ins. Co., 71 Ohio St.3d 552, 644 N.E.2d 397, 399 (1994); Nichols, 306 S.E.2d at 620. One state appears to make bad faith a strict liability tort. Hayseeds, 352 S.E.2d at 80. The other states which allow extra-contractual damages for bad faith have not specified a standard for culpable deliberateness. See Henderson I, supra, at 1158-1159; Henderson II, supra, at 40-42; see also Barker, supra, at 51.
Ill
Making the first element of the tort a legal issue rather than a factual one is also required because of the inescapable difficulties in applying traditional no-evidence review standards to bad-faith cases. Texas’ constitutional provision that courts of appeals’ decisions “shall be conclusive on all questions of fact brought before them on appeal or error”, Tex. Const, art. V, § 6, has long been construed to limit this- Court’s jurisdiction in such eases to legal questions. Choate v. San Antonio & A.P. Ry., 91 Tex. 406, 44 S.W. 69, 69 (1898). We have concluded that the absence of any significant evidence and the conclusiveness of the evidence are legal questions which we must address, but that the weight and preponderance of the evidence is a factual question within the exclusive jurisdiction of the courts of appeals. In re King’s Estate, 150 Tex. 662, 244 S.W.2d 660 (1951) (per curiam). To determine the existence of evidence to support a judgment, which we must do, without weighing the evidence, which we cannot do, we have often said that we must view the evidence in the light most favorable to the prevailing party, e.g., Moriel, 879 S.W.2d at 24; Harbin v. Seale, 461 S.W.2d 591, 592 (Tex.1970); Burt v. Lochausen, 151 Tex. 289, 249 S.W.2d 194, 199 (1952); McAfee v. Travis Gas Corp., 137 Tex. 314, 153 S.W.2d 442, 445 (1941); that we must indulge every reasonable inference in that party’s favor, Harbin, 461 S.W.2d at 592; and that we must disregard all contrary conflicting evidence and inferences, e.g., Continental Coffee Prods. Co. v. Cazarez, 937 S.W.2d 444, 450 (Tex.1996); Havner, 825 S.W.2d at 458; Orozco v. Sander, 824 S.W.2d 555, 556 (Tex.1992); Garza v. Alviar, 395 *74S.W.2d 821, 823 (Tex.1965). We do not, of course, “disregard undisputed evidence that allows of only one logical inference.” Ante at 51 n. 1 (Spector, J.); Wininger v. Ft. Worth & D.C. Ry., 105 Tex. 56, 143 S.W. 1150, 1152 (1912). See W. Wendell Hall, Revisiting Standards of Review in Civil Appeals, 24 St. MARY’S L.J. 1045, 1134 (1993); William Powers, Jr. & Jack Ratliff, Another Look at “No Evidence” and “Insufficient Evidence,” 69 Texas L.Rev. 515, 522 (1991); Robert W. Calvert, “No Evidence” and “Insufficient Evidence” Points of Error, 38 Texas L.Rev. 361, 362-63 (1960); W. St. John Garwood, The Question of Insufficient Evidence on Appeal, 30 Texas L.Rev. 803, 805 (1952).
The difficulty in applying this no-evidence standard in bad-faith cases is this. If, on the one hand, a judgment for bad faith may be supported by nothing more than the absence of evidence of a reasonable basis for denying or delaying a claim, then no judgment can be reversed for want of evidence. If all the evidence of a reasonable basis for the insurer’s actions — evidence that does not support a verdict of no reasonable basis — is disregarded, then there will never be any evidence of a reasonable basis. If, on the other hand, a judgment for bad faith must be supported by evidence negating the existence of any reasonable basis, then no judgment can survive review. No plaintiff can disprove every reasonable basis conceivable for denying or delaying a claim. Inasmuch as these are the only two alternatives — either affirm every bad-faith finding or reverse every bad-faith finding — we have quite properly referred to the problem as a logical “conundrum”. Lyons, 866 S.W.2d at 600.
Forced to choose between these two alternatives, the courts of appeals have picked the first one. I have not found a court of appeals case that applied the traditional no-evidence standard of review to a jury finding of bad faith and reversed the finding. These decisions illustrate the unlikelihood that a jury finding of bad faith will be reversed after the application of the traditional standard of no-evidence review. Stewart Title Guaranty Co. v. Aiello, 911 S.W.2d 463, 471 (Tex.App.—El Paso 1995), rev’d on other grounds, 941 S.W.2d 68 (Tex.1997); Maryland Ins. Co. v. Head Indus. Coatings & Servs., Inc., 906 S.W.2d 218, 227 (Tex.App.—Texarkana 1995), rev’d on other grounds, 938 S.W.2d 27 (Tex.1996) (per curiam); St. Paul Surplus Lines Ins. Co. v. Dal-Worth Tank Co., 917 S.W.2d 29, 55-56 (Tex.App.— Amarillo Aug. 29, 1995, writ pending); Liberty Mut. Fire Ins. Co. v. Crane, 898 S.W.2d 944, 952 (Tex.App.—Beaumont 1995, no writ); Southern Life & Health Ins. Co. v. Alfaro, 875 S.W.2d 740, 746 (Tex.App.—San Antonio 1994, no writ); Nationwide Mut. Ins. Co. v. Crowe, 857 S.W.2d 644, 652 (Tex.App.—Houston [14th Dist.]), judgm’t vacated pursuant to settlement, 863 S.W.2d 462 (Tex.1993); Shelton Ins. Agency v. St. Paul Mercury Ins. Co., 848 S.W.2d 739, 746 (Tex.App.—Corpus Christi 1993, writ denied); State Farm Fire & Cas. Co. v. Price, 845 S.W.2d 427, 438 (Tex.App.—Amarillo 1992, writ dism’d by agr.); St. Paul Ins. Co. v. Rakkar, 838 S.W.2d 622, 627 (Tex.App.—Dallas 1992, writ denied); Commonwealth Lloyd’s Ins. Co. v. Thomas, 825 S.W.2d 135, 144 (Tex.App.—Dallas 1992), judgm’t vacated pursuant to settlement, 843 S.W.2d 486 (Tex.1993); Texas Employers Ins. Ass’n v. Puckett, 822 S.W.2d 133, 138 (Tex.App.—Houston [1st Dist.] 1991, writ denied); State Farm County Mut. Ins. Co. v. Moran, 809 S.W.2d 613, 618 (Tex.App.—Corpus Christi 1991, writ denied); State Farm Mut. Auto. Ins. Co. v. Zubiate, 808 S.W.2d 590, 596 (Tex.App.—El Paso 1991, writ denied); Automobile Ins. Co. v. Davila, 805 S.W.2d 897, 906 (Tex.App.—Corpus Christi 1991, writ denied); National Union Fire Ins. Co. v. Dominguez, 793 S.W.2d 66, 70 (Tex.App.—El Paso 1990), rev’d, 873 S.W.2d 373 (Tex.1994); Wm. H. McGee & Co. v. Schick, 792 S.W.2d 513, 522 (Tex.App.— Eastland 1990), judgm’t vacated pursuant to settlement, 843 S.W.2d 473 (Tex.1992); Fidelity & Cas. Co. v. Underwood, 791 S.W.2d 635, 647 (Tex.App.—Dallas 1990, no writ); Allied Gen. Agency, Inc. v. Moody, 788 S.W.2d 601, 607 (Tex.App.—Dallas 1990, writ denied); Paramount Nat’l Life Ins. Co. v. Williams, 772 S.W.2d 255, 264 (Tex.App.—Houston [14th Dist.] 1989, writ denied); Aetna Cas. & Sur. Co. v. Joseph, 769 S.W.2d 603, 606 (Tex.App.—Dallas 1989, no writ).
*75The difficulty in applying the traditional no-evidence review standard is not restricted to bad-faith cases. The Court first noted the problem in Burk Royalty v. Walls, 616 S.W.2d 911, 922 (Tex.1981), in assessing the evidence of gross negligence, then defined as “such an entire want of care as ... shows that the act or omission was the result of conscious indifference”. (The Legislature has since changed the definition of gross negligence, bringing it under the rubric of “malice”. Tex. Civ. PRAC. & Rem.Code § 41.001(7)(B).) If all evidence of care taken by a defendant is disregarded, then every gross negligence finding must be upheld unless plaintiff must affirmatively disprove that defendant showed any care, in which event no gross negligence finding can be upheld. The Court refused to be put to this choice.
In determining whether there is some evidence of the jury’s finding of gross negligence, the reviewing court must look to all of the surrounding facts, circumstances, and conditions, not just individual elements or facts. Siebenlist v. Harville [596 S.W.2d 113, 115 (Tex.1980) ]. At first glance there may appear to be some conflict in utilizing the traditional no evidence test and considering all the facts and circumstances to determine gross negligence. [McPhearson v. Sullivan, 463 S.W.2d 174, 176 (Tex.1971), and Harbin, 461 S.W.2d at 593] indicate that the existence of gross negligence need not rest upon a single act or omission, but may result from a combination of negligent acts or omissions, and many circumstances and elements may be considered in determining whether an act constitutes gross negligence. A mental state may be inferred from actions. ■ All actions or circumstances indicating a state of mind amounting to a conscious indifference must be examined in deciding if there is some evidence of gross negligence.
“In making this determination, all evidence must be considered in a light most favorable to the party in whose favor the verdict has been rendered, and every reasonable inference dedudble from the evidence is to be indulged in such party’s favor.” Harbin [461 S.W.2d at 592], Accord: Garza [, 395 S.W.2d at 823].
616 S.W.2d at 922 (emphasis added). Four times in six sentences the Court reiterated that in a no-evidenee review of a finding of gross negligence, all the evidence must be considered, not just the evidence in favor of the finding. Specifically, viewing the evidence in the light most favorable to the gross negligence finding did not require the Court to disregard undisputed facts contrary to the finding, but only those facts as to which the evidence was conflicting. See also Moriel, 879 S.W.2d at 25 (“In evaluating legal sufficiency, we are required to determine whether the proffered evidence as a whole rises to a level that would enable reasonable and fair-minded people to differ in their conclusions.”).
The court of appeals in State Farm, Lloyds, Inc. v. Polasek, 847 S.W.2d 279 (Tex.App.—San Antonio 1992, writ denied), held that the Bwrfc-modified no-evidence review standard should be applied in bad-faith cases — that is, that all the evidence must be considered, albeit in the light most favorable to the bad-faith finding, and that undisputed evidence of a reasonable basis for denying or delaying a claim cannot be disregarded. Polasek reversed a bad-faith finding. Of the six cases that have, with varying degrees of clarity, followed the Polasek standard, five have reversed bad-faith findings. Miles, 923 S.W.2d at 806-811; Connolly v. Service Lloyds Ins. Co., 910 S.W.2d 557, 561-563 (Tex.App.—Beaumont 1995, no writ); Cortez v. Liberty Mut. Fire Ins. Co., 885 S.W.2d 466, 469-470 (Tex.App.—El Paso 1994, writ denied); Emmert v. Progressive County Mut. Ins. Co., 882 S.W.2d 32, 35-36 (Tex.App.—Tyler 1994, writ denied); Rogers v. CIGNA Ins. Co., 881 S.W.2d 177, 183 (Tex.App.—Houston [1st Dist.] 1994, no writ). Three cases prior to Polasek that applied a similar standard all reversed bad-faith findings. St. Paul Lloyd’s Ins. Co. v. Fong Chun Huang, 808 S.W.2d 524, 526 (Tex.App.—Houston [14th Dist.] 1991, writ denied); Millers Cas. Ins. Co. v. Lyons, 798 S.W.2d 339, 343-344 (Tex.App.—Eastland 1990), aff'd, 866 S.W.2d 597 (Tex.1993); National Union Fire Ins. Co. v. Hudson Energy Co., 780 S.W.2d 417, 426-427 (Tex.App.—Texarkana 1989), aff'd, 811 S.W.2d 552 (Tex.1991). Only one court applying the Polasek standard has upheld a *76finding of bad faith. Aetna Cas. & Sur. Co. v. Garza, 906 S.W.2d 548, 551 (Tex.App.—San Antonio 1995, writ dism’d).
The Polasek approach eases, but does not eliminate, the tension between the bad-faith standard of liability and the no-evidence standard of review. When there is undisputed evidence of some reasonable basis for denying or delaying a claim, Polasek’s no-evidence analysis does not require that it be disregarded. In such cases a finding of bad faith obviously cannot be supported. But in most cases the facts are disputed, and the problem of review remains. If the traditional no-evidence review standard is applied, the evidence favoring the insurer must be disregarded entirely, and the survival of a bad-faith finding will then depend on whether the insured must show an absence of evidence favoring the insurer — in which ease the finding will almost always be upheld, since all contrary evidence has been disregarded — or whether the insured must affirmatively disprove any reasonable basis for the insurer’s actions — in which case the finding will almost always fall, given the difficulty in proving a negative. In either event meaningful no-evidenee review is difficult or impossible.
In Polasek the insureds claimed benefits for the fire loss of their business. Evidence of the fire’s incendiary origins and of the insureds’ opportunity to set the fire was undisputed. The parties’ dispute was over motive. The jury could find that the insureds did not set the fire, the court of appeals concluded, but could not find that the insurer had no reasonable basis for thinking the insureds did set the fire. The evidence of motive, though inconclusive, was too strong. Given the evidence in the case, I do not disagree with the result. But it is still not clear from Polasek when evidence of no motive would have become so weak that a bad-faith finding would be found to have some evidentiary support. Making that determination when the standard of liability is no reasonable basis for denying the claim ineluctably leads to weighing the evidence, which no-evidence review, under any standard, does not permit.
We attempted to ease the dilemma in Lyons and Dominguez. In Lyons, a homeowner claimed that the brick veneer of her house was damaged in a windstorm. Her insurer concluded that the damage was caused by settling of the foundation and was therefore not covered by her policy. The homeowner sued her insurer for policy benefits and for breach of the duty of good faith and fair dealing. The jury found that the insured’s loss was covered by the policy and that the insurer had acted in bad faith. The trial court rendered judgment on the verdict. The court of appeals affirmed the judgment awarding policy benefits but reversed the award of bad-faith damages, holding that there was no evidence of bad faith. Millers Cas. Ins. Co. v. Lyons, 798 S.W.2d 389 (Tex.App.—Eastland 1990), aff'd, 866 S.W.2d 597 (Tex.1993).
Before assessing the evidence, we formulated the standard of review in bad-faith cases. We began by acknowledging that the usual no-evidence standard was not suited to bad-faith cases:
Appellate review of the legal sufficiency of the evidence supporting a judgment for the insured in a bad faith case, however, presents unusual problems. Primary among these is the conundrum of a reviewing court scouring the record to evaluate an insurer’s claim that there is “no evidence” of a negative fact, that is, that the insurer had no reasonable basis to deny or delay payment of a claim.
The traditional statement of the standard of review for legal sufficiency requires a court to consider only the evidence favoring the judgment for the insured and to disregard all evidence to the contrary. Our courts of appeals have struggled to reconcile the insurer’s substantive rights under the Aran-da test and the traditional statement of the no evidence standard of review.
866 S.W.2d at 600 (citations omitted). We then shifted focus from the no-evidence review standard to the standard of liability:
As this ease demonstrates, we believe that when a court is reviewing the legal sufficiency of the evidence supporting a bad faith finding, its focus should be on the relationship of the evidence arguably supporting the bad faith finding to the ele*77ments of bad faith. The evidence presented, viewed in the light most favorable to the prevailing party, must be such as to permit the logical inference that the insurer had no reasonable basis to delay or deny payment of the claim, and that it knew or should have known it had no reasonable basis for its actions. The evidence must relate to the tort issue of no reasonable basis for denial or delay in payment of a claim, not just to the contract issue of coverage. This is nothing more than a particularized application of our traditional no evidence review. This focus on the evidence and its relation to the elements of bad faith is necessary to maintain the distinction between a contract claim on the policy, and a claim of bad faith delay or denial of that claim, which arises from the tort duty we imposed on insurers in Arnold and Aranda.
Id. (citation omitted). Unquestionably, the evidence of coverage was conflicting. Id. at 604-05 (Doggett, J., dissenting). But a mere dispute over coverage is no evidence of bad faith. Absent evidence that Millers’ reliance on two experts’ reports was in bad faith, there was no basis for the jury’s bad-faith finding.
In Dominguez, a workers’ compensation claimant settled with the carrier for $28,000 after the claimant appealed from an Industrial Accident Board award of $6,559.48 in lump sum and lifetime medical benefits. The same day that suit was settled, the claimant sued the earner for bad faith in handling his claim. A jury awarded him $125,000 for past and future mental anguish and $175,000 exemplary damages, and the trial court rendered judgment on the verdict. The court of appeals affirmed the finding of bad faith and the award of $75,000 past mental anguish but reversed the balance of the judgment. National Union Fire Ins. Co. v. Dominguez, 793 S.W.2d 66 (Tex.App.—El Paso 1990), rev’d, 873 S.W.2d 373 (Tex.1994).
As in Lyons, we began our analysis by noting the difficulty in applying the usual no-evidence standard of review in bad-faith cases:
The claimant has the burden of proving a negative proposition, the absence of a reasonable basis for denying a claim, of which the carrier knew or should have known. As we observed in [Lyons], courts have had difficulty applying a traditional “no evidence” review of these elements of proof. The difficulty has centered around the rule of no evidence review that a court may consider only the evidence supporting the finding, disregarding all evidence and inferences to the contrary.
873 S.W.2d at 376 (citations omitted). Referring to Lyons, the Court again shifted focus from the review standard to the liability standard:
A legal sufficiency review of a bad faith claim requires that the evidence relied on by the insured as evidence of bad faith “must be such as to permit the logical inference that the insurer had no reasonable basis to delay or deny payment of the claim, and that it knew or should have known it had no reasonable basis for its actions.” Unless the evidence furnishes some reasonable basis for the conclusion by reasonable minds as to the existence of the vital fact, it is no more than a scintilla, the legal equivalent of no evidence.
As we held in Lyons, evidence of insurance coverage alone does not go to the issue of the standard articulated in Aran-da, absence of a reasonable basis for denial of the claim. A legal sufficiency analysis requires the reviewing court to give weight only to evidence supporting the judgment for the insured and reject all evidence to the contrary. However, only after an appellate court has determined what potential basis an insurance company may have had for denying a claim can the court conduct a meaningful review of whether the insured has presented evidence that the insurer lacked a reasonable basis for denying or delaying the claim. The court may then apply the traditional rules of legal sufficiency review, giving weight only to evidence in support of the judgment. “The evidence must relate to the tort issue of no reasonable basis for denial or delay in payment of a claim, not just to the contract issue of coverage.”
*78Id. (citations omitted). Thus, we held that no-evidence review could not occur until after “an appellate court determined what grounds an insurer had for showing that it acted in good faith.
Dominguez claimed that he injured his back on the job. National Union denied Dominguez’ claim as not being job-related, based on the opinion of one of the physicians who treated Dominguez that he suffered from degenerative joint disease, and Dominguez’ statements in insurance forms that his injury was not job-related. Like Lyons, Dominguez claimed that National Union acted in bad faith because it was not persuaded by other evidence, including a contrary opinion by another treating physician, that the claim was valid. The evidence of coverage was conflicting, just as in Lyons, id. at 378-79 (Doggett, J., dissenting), but that was no evidence of bad faith.
The only evidence offered by Dominguez to establish bad faith was a letter sent by a doctor to Dominguez’ attorney, stating his opinion that the injury was work related. While that letter is some evidence of coverage, it is not evidence of an absence of a reasonable basis for denying the claim. Dominguez presented no evidence that casts doubt on National Union’s reliance on the medical professionals who diagnosed the condition as a degenerative disease, or on Dominguez’ own statements on insurance forms that his condition was not work related.
Id. at 376-377 (footnote omitted). Absent evidence that National Union’s reliance on Dominguez’ own statements and the report of one of his treating physicians was in bad faith, there was no basis for the jury’s bad-faith finding.
While I agree that Lyons and Dominguez were correctly decided, neither provides a way of resolving the dilemma in no-evidenee review of bad-faith findings. Only Justioe Enooh remains of the view that Lyons is a complete solution. As the court of appeals stated in Miles, this Court “has ultimately done little to provide lower courts with any guidance for conducting a legal sufficiency review” of bad-faith findings. Miles, 923 S.W.2d at 810.
Changing the liability standard from “no reasonable basis” to “reasonably clear” certainly does not escape the difficulty of no-evidence review. Contrary to the view expressed in JUSTICE SpeotoR’s opinion, the problem is not whether the liability standard is phrased negatively or positively. The same difficulty arises, for example, in reviewing a finding of alter ego: “If a jury must consider the total dealings between a corporation and an individual before it can find alter ego, it is hardly appropriate to review the propriety of an affirmative finding by looking only to those dealings which might imply alter ego.” Mancorp, Inc. v. Culpepper, 802 S.W.2d 226, 232 n. 1 (Tex.1990) (Hecht, J., dissenting). The problem lies in applying a review standard that does not permit a weighing of the evidence to a liability standard that requires such weighing. One indication of alter ego — say the failure to keep corporate minutes up to date, for example — is no evidence of alter ego if the corporate identity is preserved in all other respects. In determining whether there is any evidence of alter ego, one cannot ignore evidence to the contrary. The issue is not whether anything was done to preserve a separate corporate identity but whether enough was done.
The same is true of bad faith, whether the standard is “no reasonable basis” or “reasonably clear”. If some evidence shows coverage and only that evidence is considered, the insurer’s obligation is reasonably clear and it has no reasonable basis to deny or delay the claim. If some evidence shows non-coverage and only that evidence is considered, the insurer’s liability is not reasonably clear and it has a reasonable basis for denying or delaying the claim. Either view makes no sense. Whether a basis for denying a claim is reasonable necessarily depends on all the evidence, just as deciding whether liability for a claim is reasonably clear. The “reasonably clear” standard does no more to avoid the difficulty with no-evidence review than the “fairly debatable” standard, which Jus-tige Spectob’s opinion rejects.
The courts of appeals’ extensive experience in no-evidence review of bad-faith cases makes it quite clear that meaningful review *79cannot be achieved by modifying the first element of bad-faith liability. An honest application of the review standard requires af-firmance of a verdict of bad faith if there is any evidence of coverage and affirmance of a verdict of no bad faith if there is any evidence of non-coverage. Monel makes clear that bad faith is more than denying a claim that is ultimately determined to have been covered. Bad faith is an unserupulousness, an arbitrariness, a taking of unfair advantage; it is not a bona fide disagreement. Only if the liability standard is a legal one can bad faith be kept separate from coverage disputes.
IV
The view of JUSTICE Spector’s opinion that no-evidence review of bad-faith liability is aided by changing the liability standard from “no reasonable basis” to “reasonably clear” can thus be shown to be incorrect. Sensing the flaws in that view, I think, Justice Speotor’s opinion attempts to bolster it with the argument that little harm will be done by the unavailability of meaningful review because the recovery of significant damages is unlikely. Specifically, Justioe Speotor’s opinion states that a “relatively stringent standard of proof ensures that punitive damages will ordinarily be available only in exceptional cases”, ante at 54, that “punitive damages in Texas bad-faith cases will be limited to highly unusual and particularly egregious situations, as they should be”, ante at 54 n. 3, and that “mental anguish damages will be limited to those cases in which the denial or delay in payment of a claim has seriously disrupted the insured’s life”, ante at 54. While I join in these views, I do not agree that making damages hard to prove is an acceptable way of correcting the flaws in the bad-faith cause of action.
The argument for creating a cause of action for insurance bad faith is that contract damages do not adequately deter insurers from unscrupulous conduct or compensate insureds for such conduct. If this is true, why create a cause of action for extra-contractual damages that can rarely be proved? Arguments for the cause of action are seriously undermined by concessions that damage awards will be rare. As a practical matter, of course, if damage awards really are rare, then the problems of defining the cause of action and providing for evidentiary review become far less important. But it remains to be seen whether the Court is serious about limiting damages for bad faith to exceptional cases, or whether the rhetoric in Justice Speotor’s opinion is merely intended to diminish the Court’s failure to rectify the problems with the bad-faith cause of action.
In any event, the validity of Justice Spec-tor’s position can now be tested in two ways: by determining whether appellate courts find evidentiary review of bad-faith judgments any easier or clearer, and whether extra-contractual damage awards are a rarity. If a decade’s experience in this State and many others is any indication, her position will not pass either test.
‡ sj: :js s|:
I agree with the other Members of the Court that Universe Life had no basis for delaying payment of Giles’ claim but that there was no evidence supporting an award of punitive damages. Accordingly, I concur in the Court’s judgment.