The opinion and judgment of the court delivered on July 13, 1983 are withdrawn, and the following is substituted.
Carolyn Parker Duncan, individually and on behalf of her minor children, brought this wrongful death action against Cessna Aircraft Company (“Cessna”) for damages suffered when an airplane crash killed her husband, James Parker.1 The jury returned a verdict of $1,000,000 for Duncan, but the trial court rendered judgment non obstante veredicto for Cessna. The court of appeals reversed the trial court’s judgment and remanded the cause for a partial new trial. 632 S.W.2d 375. We reverse the judgments of the court of appeals and the trial court and render judgment for Duncan on the jury verdict.
This case presents three questions. The first is whether Texas or New Mexico law controls the construction of a release executed by Duncan in favor of the owner of the airplane, Air Plains West, Inc. We hold that the release must be construed according to Texas law because Texas has the most significant relationship to this issue.
The second question is whether, under Texas law, the release discharged Cessna’s liability to Duncan. We hold that Cessna was not discharged because it was not specifically identified in the release.
*418The final question before us is whether Cessna, a strictly liable manufacturer, is entitled to contribution from Smithson’s estate based on proof that his pilot negligence caused the fatal crash. We hold that in products liability cases tried after July 13, 1983, the date of our former opinion, a defendant may obtain a jury allocation of the plaintiffs damages according to the plaintiff’s, defendants’, and third parties’ respective percentages of causation of those damages. We also hold, however, that Cessna did not preserve its claim for contribution against Smithson’s estate.
I. BACKGROUND
Benjamin Smithson and James Parker died in the crash of a Cessna 150 airplane in New Mexico in 1976. At the time of the crash, Smithson was employed as an instructor pilot for Air Plains West, Inc., which owned the airplane, and was giving Parker flying lessons.
Parker’s widow, Carolyn Duncan, individually and on behalf of her minor children, filed a wrongful death action in Federal District Court for the Northern District of Texas against Air Plains West and Smithson’s estate. She alleged that their negligence proximately caused the crash and her husband’s death. The suit was terminated when Duncan settled with Air Plains West for $90,000 and executed a release (“Duncan release”) that stated, in pertinent part,
we [Duncan and her minor children] ... do hereby release, discharge and forever quitclaim Air Plains West, Inc., its agents, servants and employees, and the Estate of Benjamin A. Smithson, Jr., deceased, or any other corporations or persons whomsoever responsible therefor, whether named herein or not, from any and all claims of every kind and character whatsoever, and from any cause of action, claims, demands, costs, loss of services, compensation, medical, hospital and doctor expenses, funeral and burial expenses, and damages, both actual and exemplary, on account of the fatal injuries sustained by the said James E. Parker, which resulted in his death, as the result of an airplane crash occurring on or about October 19, 1976 _ (emphasis added).
Duncan and Mrs. Smithson subsequently instituted wrongful death actions against Cessna.2 They alleged that design and manufacturing defects in the legs of the cockpit seats caused the legs to break during the crash, causing the deaths of their husbands.
Cessna responded with a counterclaim against Smithson’s estate. The counterclaim asserted that Smithson’s negligence had caused the crash and that Cessna was therefore entitled to contribution from his estate for any damages Duncan recovered from Cessna. Mrs. Smithson specially excepted to the counterclaim on the ground that her husband’s estate was entitled to full indemnification from Cessna for damages caused by her husband’s negligence, if any. The trial court sustained the special exception and struck Cessna’s counterclaim from the lawsuit.
In addition, in its first amended original answer to Duncan’s petition, Cessna claimed that its liability to the Duncan family was discharged by the Duncan release of Air Plains West. Cessna again alleged that Smithson’s negligence caused the crash. Based on that allegation and Duncan’s settlement agreement with Air Plains West and Mr. Smithson’s estate, Cessna alternatively sought a one-half reduction of any damages Duncan recovered from Cessna. See Palestine Contractors, Inc. v. Perkins, 386 S.W.2d 764 (Tex.1964). Duncan’s special exception to this allegation urged that the trial court’s previous decision to strike Cessna’s counterclaim precluded any pleading alleging Smithson’s negligence.
Cessna then filed a second amended original answer. Like Cessna’s first amended *419original answer, it included an allegation of Smithson’s negligence and a claim for a Palestine Contractors reduction. Duncan did not specially except to Cessna’s second amended original answer. Before the trial concluded, Cessna offered deposition evidence pertaining to Smithson’s negligence in causing the crash. Although the court excluded the evidence, Cessna did not attempt to make this putative deposition a part of the record through a formal bill of exceptions. See TEX.R.CIY.P. 372.
The trial court did not rule on Duncan’s special exceptions to Cessna’s first amended original answer until after the jury returned its verdict. At that time, the trial court finally signed an order sustaining those special exceptions. The court then granted Cessna’s motion for judgment non obstante veredicto on the ground that the release executed in favor of Air Plains West and Smithson’s estate also discharged Cessna’s liability.
II. THE CONFLICTS PROBLEM
Cessna argues that in determining the effect of the Duncan release on Cessna’s liability, we should apply New Mexico law, not Texas law. Cessna further argues that the New Mexico courts would construe the Duncan release to bar Duncan’s cause of action against Cessna for damages arising out of the plane crash.
Duncan, on the other hand, contends that this case presents no true conflicts problem. She argues that we do not need to decide which state’s law applies because under either Texas or New Mexico law, the general language in her release did not discharge Cessna. In order to resolve the effect of the release, therefore, we must first determine whether there is a difference between the rules of Texas and New Mexico on this issue.
A. Effect of the Release Under Texas Law
In McMillen v. Klingensmith, 467 S.W.2d 193 (Tex.1971), this court abolished the common law “unity of release” rule, under which a release of one joint tort-feasor fully discharged all remaining tort-feasors. We held that a release fully discharges only those tortfeasors that it names or otherwise specifically identifies. Id. at 196.
In this case, the Duncan release purports to discharge “any other corporations or persons whomsoever responsible” for the death of James Parker. Cessna clearly falls within this general class of tort-feasors. The question, therefore, is whether the naming of a general class of tort-feasors constitutes specific identification of each member of the class.
The courts confronting this question after McMillen have reached contrary results. In Bell v. First National Bank, 597 S.W.2d 521 (Tex.Civ.App.—Dallas 1980, no writ), a car owner sued a bank for wrongful repossession after releasing the dealer who also participated in the repossession. The court held that the general reference in the release to “all other persons, firms, or corporations ... in any way connected with [the tortious event]” reflected a clear intent to specifically identify anyone incurring liability because of the repossession. Id. at 522. As a result, the release barred the suit against the bank.
In Lloyd v. Ray, 606 S.W.2d 545, 547 (Tex.Civ.App.—San Antonio 1980, writ ref’d n.r.e.), however, the court held that a release of a malpractice claim against a physician and “all other persons, firms, and corporations” did not specifically identify, and therefore did not discharge, a second physician. Accord Duke v. Brookshire Grocery Co., 568 S.W.2d 470, 472 (Tex.Civ.App.—Texarkana 1978, no writ).
The court of appeals in this case held that the McMillen requirement of specific identification is not met unless the reference in the release is so particular that “a stranger could readily identify the released party.” 632 S.W.2d at 381. We agree. Accordingly, we approve Lloyd and Duke and disapprove Bell. We hold that under Texas law, the mere naming of a general class of tortfeasors in a release does not discharge the liability of each *420member of that class. A tortfeasor can claim the protection of a release only if the release refers to him by name or with such descriptive particularity that his identity or his connection with the tortious event is not in doubt. In this case, the release does not name Cessna, nor does it provide some specific description of Cessna. Since the reference to “all corporations” does not supply the descriptive particularity necessary to specifically identify Cessna, the release does not bar Duncan’s action if Texas law applies to its construction.
B. Effect of the Release Under New Mexico Law
Section 4 of New Mexico’s version of the Uniform Contribution Among Tortfeasors Act provides, “A release by the injured party of one joint tortfeasor, whether before or after judgment, does not discharge the other tortfeasor unless the release so ■provides.” N.M.Stat.Ann. § 41-3-4 (emphasis added).
In Johnson v. City of Las Cruces, 86 N.M. 196, 521 P.2d 1037 (Ct.App.1974), a New Mexico Court of Appeals interpreted this statute and held that a release discharges any tortfeasor who is named or who comes within a named general class. The court said that the releasor’s failure to allege any ambiguity in the releasing instrument precluded any contention that all tortfeasors were not intended to be discharged. Although the New Mexico Supreme Court did not approve or disapprove this holding, we must assume that Johnson correctly states the applicable rule in New Mexico.3 Because Johnson is contrary to our decision in McMillen v. Klingensmith, we must decide whether Texas or New Mexico law applies in construing the release.
C. The Law Applicable to Construction of the Release
One week before trial, Cessna moved the court to take judicial notice of certain statutes and case law of New Mexico. See TEX.R.CIV.P. 184a. The court granted Cessna’s motion and ruled that New Mexico law, to the extent that it differed from Texas law, applied to all substantive issues in the case. The court of appeals, however, held that Texas law governed the construction of the release because the release was a contract executed in Texas and, in accordance with the rule of lex loci contractus, the law of the place of the making of the contract applied to its construction.
Cessna and Duncan both argue that the court of appeals erred in applying the rule of lex loci contractus. They contend that the correct approach is the most significant relationship methodology of the Restatement (Second) of Conflict of Laws, which we adopted in Gutierrez v. Collins, 583 S.W.2d 312 (Tex.1979), for tort choice of law issues.4
In Gutierrez, this court strongly criticized the traditional tort choice of law rule of lex loci delecti (i.e., apply the law of the place where the wrong occurred). We decided that the rule’s ease of application and uniformity of result did not justify its arbitrary and often inequitable results. Additionally, we noted that the traditional rule did not meet the demands of our highly mobile modern society. Primarily for these reasons, we abandoned lex loci delecti and replaced it with the most significant relationship approach set forth in §§ 65 and *421145 of the Restatement (Second) of Conflict of Laws. The significant relationship methodology was selected because it offers a “rational yet flexible approach to conflicts problems, ... represents a collection of the best thinking on this subject ... [and] include[s] 'most of the substance’ of all the modern theories.” Id. at 318.
Most of the numerous inadequacies inherent in lex loci delecti also exist in the other traditional lex loci rules, including lex loci contractus. Each of these traditional rules sacrifices just and reasoned results for the ease and predictability of mechanistic decision making. However, “[e]ase of administration alone is a wholly inadequate reason for retention of an unjust rule.” Id. at 317. By contrast, use of the most significant relationship approach in accordance with the general principles stated in § 6 produces reasoned choice of law decisions grounded in those specific governmental policies relevant to the particular substantive issue. Consequently, the lex loci rules will no longer be used in this state to resolve conflicts problems. Instead, in all choice of law cases, except those contract cases in which the parties have agreed to a valid choice of law clause, the law of the state with the most significant relationship to the particular substantive issue will be applied to resolve that issue.
In applying § 6 to this case, we must first identify the state contacts that should be considered. Once these contacts are established, the question of which state’s law will apply is one of law. Gutierrez, 583 S.W.2d at 319. Moreover, the number of contacts with a particular state is not determinative. Some contacts are more important than others because they implicate state policies underlying the particular substantive issue. Consequently, selection of the applicable law depends on the qualitative nature of the particular contacts. Id. at 319.
The following undisputed state contacts are present in this case: Cessna is a Kansas corporation; the plane’s defective seats were designed and manufactured in Kansas; the plane was put into the stream of commerce in Texas; decedent James Parker lived in Texas and worked in New Mexico; decedent Benjamin Smithson lived and worked in New Mexico; Air Plains West is a New Mexico corporation; and the release was executed in Texas as a settlement of a lawsuit filed in a federal district court in Texas. The beginning point for evaluating these contacts is the identification of the policies or “governmental interests,” if any, of each state in the application of its rule.
Since under New Mexico law, the Duncan release would discharge Cessna, New Mexico has no governmental interest in the resolution of this issue. Its rule, set forth in Johnson, reflects policies of effectuating the intent of the parties to the release and of protecting New Mexico defendants. Here, however, no New Mexico defendant or injured party is involved. Cessna is a Kansas corporation seeking the benefit of a release executed in Texas by an injured Texas resident as part of a settlement of a suit filed in Texas. We can conceive of no legitimate reason why the New Mexico legislature should be concerned with the application of its statute to a Texas settlement to cut off a Texas resident’s claim against a Kansas corporation.6
*422Texas, on the other hand, has direct and important interests in the effect given to the Duncan release. Our purposes in abolishing the “unity of release” rule and requiring specific identification' of nonsettling tortfeasors were to encourage both partial and full settlements; to avoid unfairly depriving injured Texas residents of their full satisfactions; to avoid providing an incentive to tortfeasors to delay or to stay out of early settlement negotiations; and to ensure that Texas claimants do not inadvertently lose their valuable rights against unnamed and perhaps unknown tortfeasors by settling with only one of the wrongdoers. See Knutson v. Morton Foods, Inc., 603 S.W.2d 805, 807-08 (Tex.1980); McMillen v. Klingensmith, 467 S.W.2d at 195-96. To effectuate these policies, we must narrowly construe general, categorical release clauses. A nonset-tling tortfeasor should not fortuitously escape compensating his Texas victim simply because of settlement arrangements that did not encompass him or his conduct and to which he contributed nothing. If Cessna obtains the benefit of the Duncan release, then the policies underlying McMillen will obviously be frustrated.
Texas has an additional interest. Duncan could reasonably have expected that the laws of Texas would govern the effect on third parties of a settlement agreement negotiated and executed in Texas. She certainly had no reason to believe that New Mexico law would apply to cut off her cause of action against Cessna or some other unknown tortfeasor. Texas has an interest in protecting Duncan’s reasonable contractual expectations.
An analysis of the relevant state contacts reveals that New Mexico has no underlying interest in the application of its law, while Texas has important interests in allowing Duncan’s action against Cessna. In this situation, known as a “false conflict,” it is an established tenet of modern conflicts law that the law of the interested state should apply. See, e.g., J. Martin, Perspectives on Conflict of Laws: Choice of Law 85 (1980); E. Scoles & P. Hay, Conflict of Laws § 2.6, at 17 (1982); R. Weintraub, Commentary on the Conflict of Laws § 3.1, at 48, § 2.6, at 267-68 (2d ed. 1980). Accordingly, we hold that Texas law applies to the determination of the effect of the Duncan release on Cessna. We further hold that Cessna’s liability to Duncan is not discharged.
III. CESSNA’S CONTRIBUTION CLAIM AGAINST SMITHSON’S ESTATE
Cessna argues that it is entitled to contribution from Smithson’s estate for part of Duncan’s damages because Smithson’s pilot negligence proximately caused the fatal crash. The court of appeals agreed and remanded the cause for a partial retrial to allow Cessna to adduce proof of Smithson’s negligence. In this court, Cessna claims that on remand its liability to Duncan should be reduced in accordance with principles of comparative apportionment. Because Cessna also seeks a similar reduction in its liability to Smithson’s estate, we must address two distinct, but closely related questions: (1) whether a plaintiff’s contributory negligence is a defense in strict products liability actions when that negligence does not rise to the level of assumed risk or unforeseeable product misuse, but is more than a mere failure to discover a product defect; and (2) whether comparative contribution among tortfeasors is possible when at least one is strictly liable.
A. Apportioning Liability Under Existing Texas Law
In Shamrock Fuel & Oil Sales Co. v. Tunks, 416 S.W.2d 779 (Tex.1967), and McKisson v. Sales Affiliates, Inc., 416 S.W.2d 787 (Tex.1967), this court adopted section 402A of the Restatement (Second) of Torts in recognizing a cause of action based on strict products liability. We held *423that a plaintiff’s contributory negligence was no defense to this cause of action. We based this holding on comment “n” of section 402A, which states,
Contributory negligence of the plaintiff is not a defense when such negligence consists merely in a failure to discover the defect in a product or guard against its existence. On the other hand the form of contributory negligence which consists in voluntarily and unreasonably proceeding to encounter a known danger, and commonly passes under the name of assumption of risk, is a defense under this Section as in other eases of strict liability.
In Henderson v. Ford Motor Co., 519 S.W.2d 87 (Tex.1974), we reaffirmed our adherence to comment “n”, stating, “Assumption of risk is the defense; contributory negligence or failure to act reasonably is not.” Id. at 91; see also Rourke v. Garza, 530 S.W.2d 794, 800 (Tex.1975).
In General Motors Corp. v. Hopkins, 548 S.W.2d 344 (Tex.1977), however, we recognized that a manufacturer is not an insurer and “should not be required to pay for all of the damages suffered by a product user who contributed to the cause of his harm by unforeseeable handling of the product.” Id. at 351. Thus, for cases involving unforeseeable product misuse, we adopted a scheme of comparative causation under which a manufacturer’s liability is limited to the proportionate harm caused by its product. Id. at 352.
Under present Texas law, then, the only defenses to a strict product liability action are the absolute defense of assumption of the risk and the comparative defense of unforeseeable product misuse. By contrast, in product liability actions based on negligence and breach of warranty, these defenses are subsumed within contributory negligence, which operates as a comparative defense. See Signal Oil & Gas Co. v. Universal Oil Products, 572 S.W.2d 320 (Tex.1978); Tex.Rev.Civ.Stat. Ann. art. 2212a, § 1. “The procedural problems generated by this state of the substantive law are apparent.” Pope & Lowerre, The State of the Special Verdict —1979, 11 St. Mary’s L.J. 47, 48 (1979). In Texas, a plaintiff can predicate a product liability action on one or more of at least three theories of recovery: (1) strict liability under § 402A, (2) breach of warranty under the U.C.C., and (3) negligence. If either breach of warranty or negligence is alleged in addition to strict liability, the product supplier is entitled to submit three defensive issues, namely, contributory negligence, assumption of the risk, and misuse. In addition, he may also be entitled to submission of both a comparative negligence issue and a comparative causation (product misuse) issue.
If there were any qualitative difference between contributory negligence, on the one hand, and assumed risk and misuse, on the other, this procedural complexity might be justified. Assumed risk and unforeseeable misuse, however, are nothing more than extreme variants of contributory negligence. To varying degrees, all three defenses focus on the reasonableness of a plaintiff’s conduct. Boatland of Houston, Inc. v. Bailey, 609 S.W.2d 743, 750-51 (Tex.1980) (Pope, J., concurring); see also, e.g., Farley v. M M Cattle Co., 529 S.W.2d 751, 758 (Tex.1975).
A similar problem has confronted courts attempting to apportion joint liability among negligent and strictly liable tort-feasors. Texas has two statutes that allow contribution among tortfeasors. Article 2212a provides for comparative contribution among joint tortfeasors in negligence cases according to their respective percentages of fault. Article 2212, on the other hand, mandates pro rata contribution from co-tortfeasors in other tort actions.
In General Motors Corp. v. Simmons, 558 S.W.2d 855 (Tex.1977), we stated that article 2212a is inapplicable to cases in which a strict liability claim is asserted against at least one defendant. We concluded that strict liability under § 402A of the Restatement (Second) of Torts is not based on negligence and that article 2212 governs contribution among joint tort-feasors when one of them is strictly liable. *424We also observed, however, that the apparent failure of articles 2212 and 2212a to fairly and sensibly compare causation in such situations promised to engender serious problems in future products liability cases. For this reason, we invited legislative study and corrective action.
In recent years, products liability litigation has spawned so many intractable problems of loss allocation between negligent plaintiffs and negligent and strictly liable defendants that our brief expression of concern in Simmons now seems understated. See, e.g., Wenzel v. Rollins Motor Company, 598 S.W.2d 895 (Tex.Civ.App.—El Paso 1980, writ ref d n.r.e.); Bell Helicopter v. Bradshaw, 594 S.W.2d 519 (Tex.Civ.App.—Corpus Christi 1979, writ ref’d n.r.e.); Avery v. Maremont Corp., 628 F.2d 441 (5th Cir.1980) (applying Texas law); Foster v. Ford Motor Company, 616 F.2d 1304 (5th Cir.1980) (applying Texas law). Simmons itself has prevented apportionment of liability in at least one case. See Pyramid Derrick & Equipment Co. v. Mason, 617 S.W.2d 727, 728-29 (Tex.Civ.App.-Beaumont 1981, writ ref’d n.r.e.). Courts have had to wrestle with various problematic indemnity doctrines and to recognize “shadowy distinctions between defenses in products cases and negligence cases.” Thus, manufacturers have often borne accident costs generated in part by the substandard conduct of the plaintiff or some third party. Boatland of Houston, Inc. v. Bailey, 609 S.W.2d at 750, 751 (Pope, J., concurring).
Product liability suits which are not based on negligence are treated anomalously because the system of allocation provided in Art. 2212a does not apply, and there is no other comprehensive system for loss allocation under existing law. This court has previously attempted to ameliorate the harsh and inequitable consequences stemming from this anomalous treatment of loss allocation in products liability actions. Thus, instead of recognizing “all or nothing” issues in product misuse and breach of implied warranty cases, we created comparative apportionment schemes. See General Motors Corp. v. Hopkins, 548 S.W.2d at 351-52 (pure comparative causation); Signal Oil & Gas Co. v. Universal Oil Products, 572 S.W.2d at 329 (pure comparative negligence). Nevertheless, our limited efforts in Hopkins and Signal Oil, though positive, have not substantially alleviated the intolerable confusion, unmanageability, and inherent unfairness in this area of Texas products law. See Boatland of Houston, Inc. v. Bailey, 609 S.W.2d at 751 (Pope, J., concurring). As a result, we have been implored repeatedly to recognize some form of comparative fault in strict products liability actions. See Edgar, Products Liability in Texas, 11 Tex.Tech L.Rev. 23, 50 (1979); Sales, Contribution and Indemnity Between Negligent and Strictly Liable Tortfeasors, 12 St. Mary’s L.J. 323, 363-65 (1980); Special Project, Texas Tort Law in Transition, 57 Texas L.Rev. 381, 491-95 (1979). This case presents us with the opportunity to consider whether we can or should adopt such a system.
B. Judicial Trend Toward Comparative “Fault”
A significant majority of the numerous commentators addressing the question have strenuously urged the implementation of comparative fault, also referred to as comparative responsibility or comparative causation, as a means of distributing accident costs among negligent plaintiffs, negligent defendants, and strictly liable defendants. See, e.g., Edgar, supra, at 44-50; Sales, supra, at 357-65; Schwartz, Strict Liability and Contributory Negligence, 42 Tenn.L.Rev. 171, 178-81 (1974); Wade, Products Liability and Plaintiffs Fault — The Uniform Comparative Fault Act, 29 Mercer L.Rev. 373, 376-81 (1978). They have pointed out on the one hand that strict products liability is not absolute liability — that is, product suppliers are not insurers of the safety of their products. On the other hand, “all or nothing” strict liability defenses are outmoded and undesirable doctrinal throwbacks resulting in unfairness to plaintiffs, to defendants, and to other product purchasers who ultimately *425absorb the loss through price setting. See, e.g., Sales, Assumption of the Risk and Misuse in Strict Tort Liability — Prelude to Comparative Fault, 11 Tex.Tech L.Rev. 729, 776-78 (1980); Special Project, supra, at 485-94. In the absence of apportionment, some manufacturers bear the total expense of accidents for which others are partly to blame, while other manufacturers totally escape liability even though they have sold defective products. Either result is unacceptable.
Unfairness, however, is not the only serious flaw of virtually ignoring plaintiff and third party misconduct in strict products liability actions. The failure to allocate accident costs in proportion to the parties’ relative abilities to prevent or to reduce those costs is economically inefficient. Special Project, supra, at 485-86. An ideal tort system should impose responsibility on the parties according to their abilities to prevent the harm. Existing law, however, encourages manufacturers to make safety improvements that are not cost justified, while failing to deter the substandard conduct of other tortfeasors. Id. Thus, equitable and rational risk distribution, a fundamental policy underlying the imposition of strict products liability, logically depends on the existence of some system for comparing causation in cases involving plaintiff or third party misconduct.
For these reasons, most of the courts addressing the issue have decided to adopt some form of comparative causation for strict liability in tort. These decisions have often been preceded by extensive discussions of the procedural and policy benefits of proportionate loss allocation.7 See, e.g., Pan-Alaska Fisheries, Inc. v. Marine Construction & Design Co., 565 F.2d 1129 (9th Cir.1977) (Federal admiralty jurisdiction); Butaud v. Suburban Marine & Sporting Goods, Inc., 555 P.2d 42 (Alaska 1976); Safeway Stores, Inc. v. Nest-Kart, 21 Cal.3d 322, 579 P.2d 441, 146 Cal.Rptr. 550 (1978); Daly v. General Motors Corp., 20 Cal.3d 725, 575 P.2d 1162, 144 Cal.Rptr. 380 (1978); West v. Caterpillar Tractor Co., 336 So.2d 80 (Fla.1976); Kaneko v. Hilo Coast Processing, Haw., 654 P.2d 343 (1982); Trust Corp. v. Piper Aircraft Corp., 506 F.Supp. 1093 (D.Mont.1981); Sandford v. Chevrolet Division of General Motors, 292 Or. 590, 642 P.2d 624 (1982); Murray v. Fairbanks Morse, 610 F.2d 149 (3d Cir.1979) (Virgin Islands); Star Furniture Co. v. Pulaski Furniture Co., 297 S.E.2d 854 (W.Va.1982); Dippel v. Sciano, 37 Wis.2d 443, 155 N.W.2d 55 (1967). We agree with these courts that applying principles of comparative apportionment to strict products liability not only furthers the policy goals of § 402A, but also simplifies the submission of products cases.
Courts applying comparative fault in strict liability actions have taken several valid approaches. In Daly v. General Motors Corp., 20 Cal.3d 725, 575 P.2d 1162, 144 Cal.Rptr. 380 (1978), the California Supreme Court, unencumbered by a comparative negligence statute, extended its judicially formulated comparative negligence system to actions involving both negligence and strict liability claims because “logic, justice, and fundamental fairness” required it. Id. 575 P.2d at 1172, 144 Cal.Rptr. at 390. The court observed that “the fundamental and underlying purpose of [our previous decision to adopt comparative negligence] was to promote the equitable allocation of loss among all parties legally responsible in proportion to their fault.” 575 P.2d at 1169, 144 Cal.Rptr. at 387. Other jurisdictions with judicially formulated comparative negligence have also embraced comparative fault for products cases. See, e.g., Edwards v. Sears, Roebuck & Co., 512 F.2d 276 (5th Cir.1975) (applying Mississippi law); Butaud v. Suburban Marine & Sporting Goods, Inc., 555 P.2d 42 (Alaska *4261976); West v. Caterpillar Tractor Co., 336 So.2d 80, 90 (Fla.1976); cf. Scott v. Rizzo, 96 N.M. 682, 634 P.2d 1234, 1240 (1981) (reserving judgment on the question, but favorably discussing allocation in strict liability cases). In addition, one Texas court has praised the fairness and procedural simplification achieved by the Daly rule. See Wenzel v. Rollins Motor Co., 598 S.W.2d 895, 901 (Tex.Civ.App.—El Paso 1980, writ ref d n.r.e.).
Several other courts have interpreted comparative negligence statutes as encompassing strict liability in tort. In the leading case of Dippel v. Sciano, 37 Wis.2d 443, 155 N.W.2d 55 (1967), for example, the Wisconsin Supreme Court characterized strict liability as negligence per se, or negligence as a matter of law, because it arises from a violation of a standard of safety and requires no showing of foreseeability of harm. It followed, then, that the Wisconsin comparative negligence statute must apply to strict liability as well. Minnesota has adopted the “Wisconsin rule.” See Busch v. Busch Construction Co., 262 N.W.2d 377, 393 (Minn.1977); see also Jensvold, A Modern Approach to Loss Allocation Among Tortfeasors in Products Liability Cases, 58 Minn.L.Rev. 723, 748-51 (1974); Uniform Comparative Fault Act § 1 comment (1977).
In Suter v. San Angelo Foundry & Machine Company, 81 N.J. 150, 406 A.2d 140, 145-46 (1979), the New Jersey Supreme Court also applied its comparative negligence statute to strict liability in tort. Instead of characterizing strict liability as negligence per se, however, the court determined that the New Jersey legislature intended to allow the apportioning of accident costs according to “fault,” a concept subsuming negligence. Id. at 145. One reason given for this conclusion was that New Jersey adopted the provisions of the Wisconsin comparative negligence statute after the Wisconsin Supreme Court in Dip-pel had already applied its statute to strict liability. The court then observed that distributing an unsafe product is a departure from a required standard of conduct — in other words, fault. Accordingly, the court held that the statute required the comparison of fault between negligent and strictly liable parties. Other courts have used similar reasoning in applying comparative negligence statutes to strict liability actions. See Sun Valley Airlines, Inc. v. Avco-Lycoming Corp., 411 F.Supp. 598, 603 (D.Idaho 1976); Kennedy v. City of Sawyer, 228 Kan. 439, 618 P.2d 788, 796-97 (1980); Baccelleri v. Hyster Company, 287 Or. 3, 597 P.2d 351, 354-55 (1979).
Finally, some courts in jurisdictions with comparative negligence statutes have declined to interpret those statutes to encompass strict products liability cases, but have nevertheless judicially adopted separate comparative causation systems for such cases. In Thibault v. Sears, Roebuck & Co., 118 N.H. 802, 395 A.2d 843 (1978), for instance, the New Hampshire Supreme Court held that its comparative negligence statute did not apply to strict liability cases because the statute referred only to actions for negligence, but “judicially recognize[d] the comparative concept in strict liability cases parallel to the legislature’s recognition of it in the area of negligence.” 395 A.2d at 848, 850. See also Stueve v. American Honda Motors, 457 F.Supp. 740, 751-56 (D.Kan.1978) (predicting that the Kansas Supreme Court would not apply its comparative negligence statute, but would probably instead adopt a comparative system in strict liability cases as a principle of common law).
We recognize that because strict liability is closely analogous to negligence per se, our comparative negligence statute, article 2212a, might reasonably be interpreted to include both types of actions. See Dippel v. Sciano, 155 N.W.2d at 64-65. However, article 2212a refers only to negligence actions, even though strict products liability had been judicially adopted six years earlier. See McKisson v. Sales Affiliates, 416 S.W.2d 787, 788-89 (Tex.1967). Consequently, we reaffirm our observation in General Motors Corp. v. Simmons, 558 S.W.2d 855 (Tex.1977), that article 2212a does not apply to actions in which strict liability is established.
*427The legislature’s decision to limit Art. 2212a to negligence cases does not, however, preclude this court from fashioning a common law comparative apportionment system for strict products liability cases. See V. Schwartz, Comparative Negligence § 12.1, at 196 (1974); Wade, Products Liability and Plaintiffs Fault, 29 Mercer L.Rev. 373, 380 (1978); see also Stueve v. American Honda Motors, 457 F.Supp. at 751; Thibault v. Sears, Roebuck & Co., 395 A.2d at 848. In fact, the existence of a comparative negligence statute is a persuasive reason why this court should recognize such a system as a sound principle for our developing common law. See V. Schwartz, supra (citing Stone, The Common Law in the United States, 50 Harv.L. Rev. 4 (1936)). “It has always been the duty of the common-law court to perceive the impact of major legislative innovations and to interweave the new legislative policies with the inherited body of common-law principles.” Moragne v. States Marine Lines, Inc., 398 U.S. 375, 392, 90 S.Ct. 1772, 1783, 26 L.Ed.2d 339 (1970) (creating an action for wrongful death under federal maritime law).
The older contribution statute, Art. 2212, likewise does not preclude judicial adoption of comparative loss allocation. In General Motors Corp. v. Simmons we held Art. 2212 applicable to a case involving both negligent and strictly liable defendants. Article 2212 was enacted in 1917, long before any identifiable body of strict products liability law existed. It is unlikely that the legislature intended for the statute to control in cases based the unforeseen theories of products liability, presenting unforeseen questions of loss allocation. It thus appears that our holding in Simmons was not mandated by the legislative purpose supporting Art. 2212. As a result, Simmons is an unjustified obstacle to the development of a rational and fair system of allocating losses among negligent plaintiffs, negligent defendants and strictly liable defendants. Therefore, we hold that Art. 2212 does not apply when strict products liability, including uncrash-worthiness and breach of warranty, is established. To the extent it is inconsistent with our rejection of Art. 2212, General Motors Corp. v. Simmons is overruled.
We agree with the analysis of the New Hampshire court in Thibault v. Sears. Judicial adoption of a comparative apportionment system, independent of statutory comparative negligence, is a feasible and desirable means of eliminating confusion and achieving efficient loss allocation in strict liability cases. Accordingly, we hereby adopt such a system.
Many courts and commentators have labeled this type of loss allocation system comparative fault. We choose comparative causation instead because it is conceptually accurate in cases based on strict liability and breach of warranty theories in which the defendant’s “fault,” in the traditional sense of culpability, is not at issue. The trier of fact is to compare the harm caused by the defective product with the harm caused by the negligence of the other defendants, any settling tortfeasors and the plaintiff.8 The fault or conduct of the products defendant is not at issue. This important point was expressly recognized by both the majority and dissenting opinions in Lewis v. Timco, Inc., 716 F.2d 1425 (5th Cir.1983) (en banc). See also Pan-Alaska Fisheries, Inc. v. Marine Construction & Design Co., 565 F.2d at 1139; *428Butaud v. Suburban Marine & Sporting Goods, Inc., 555 P.2d at 47 (Rabinowitz, J., concurring); Daly v. General Motors Corp., 20 Cal.3d at 736, 575 P.2d at 1168, 144 Cal.Rptr. at 386. Our decision to adopt the term “comparative causation” is consistent with our focus on causation in Signal Oil & Gas Co. v. Universal Oil Products, 572 S.W.2d 320 (Tex.1978) and General Motors Corp. v. Hopkins, 548 S.W.2d 344 (Tex.1977). Under comparative causation,
Plaintiffs will continue to be relieved of proving that the manufacturer or distributor was negligent in the production, design, or dissemination of the article in question. Defendant’s liability for injuries caused by a defective product remains strict.
Daly v. General Motors Corp., 20 Cal.3d at 736-37, 575 P.2d at 1168, 144 Cal.Rptr. at 386. As a result, the product supplier’s incentive to eliminate or to reduce product hazards should remain intact. Id. at 737-38, 575 P.2d at 1169, 144 Cal.Rptr. at 387.
In cases not controlled by Art. 2212a, the system we adopt will allow comparison of plaintiff’s conduct, whether it is characterized as assumption of risk, misuse, or failure to mitigate or avoid damages, with the conduct or product of a defendant, whether the suit combines crashworthiness or other theories of strict products liability, breach of warranty, or negligence. Cf. Uniform Comparative Fault Act § 1 (1977) (proposing similar statutory system). Assumption of risk and misuse will no longer be separate defenses, but will be subsumed under the more familiar notion of contributory negligence. Accordingly, to the extent that they are inconsistent with this opinion, Henderson v. Ford Motor Co., 519 S.W.2d 87 (Tex.1974) and General Motors Corp. v. Hopkins, 548 S.W.2d 344 (Tex.1977) are overruled.
Comparative causation is especially appropriate in crashworthiness cases where the product defect causes or enhances injuries but does not cause the accident. The conduct which actually causes the accident, on the other hand, would not cause the same degree of harm if there were no product defect. Rather, it is a combination of factors that causes plaintiff’s injuries. The jury is asked to apportion responsibility between all whose action or products combined to cause the entirety of the plaintiff’s injuries. Steuve v. American Honda Motor Co., 457 F.Supp. 740, 759-60 (D.Kan.1978); see also Fietzer v. Ford Motor Co., 590 F.2d 215 (7th Cir.1978) (comparing causation in automobile crashworthiness case); Trust Corp. of Montana v. Piper Aircraft Corp., 506 F.Supp. 1093, 1098-99 (D.Mont.1981) (comparing fault in airplane crash-worthiness case); Daly v. General Motors Corp., 20 Cal.3d 725, 575 P.2d 1162, 144 Cal.Rptr. 380 (1978) (comparing fault in automobile crashworthiness case).
C. Pure Versus Modified Comparison
We must now decide, however, whether to adopt modified or “pure” comparative apportionment. Under a modified system, as exemplified by article 2212a, a claimant’s recovery is entirely barred if his share of causation is found to be greater than the total causation attributed to the defendants. Under a “pure” system, on the other hand, a claimant can recover the percentage of damages caused by the defendants, regardless of the extent of his own causation.
Most authorities agree that a pure comparative scheme is preferable to a modified one and that when possible, courts should allow pure comparison, especially in products cases. See, e.g., V. Schwartz, supra §§ 12.6, 12.7 at 207, 209; Prosser, Comparative Negligence, 51 Mich.L.Rev. 465, 494 (1953); Special Project, supra at 421-24. Moreover, when not restricted by statutes, courts almost invariably opt for pure comparison because they recognize the relative inefficiency and reduced deterrent effect of modified comparative apportionment. Special Project, supra at 424; see also Daly v. General Motors Corp., 575 P.2d at 1168-69, 144 Cal.Rptr. at 386-87. *429On two prior occasions, this court has chosen pure comparison in products cases. Signal Oil & Gas Co. v. Universal Oil Products, 572 S.W.2d 320 (Tex.1978) (breach of implied warranty actions); General Motors Corp. v. Hopkins, 548 S.W.2d 844 (Tex.1977) (product misuse).
In this case, the policies underlying strict liability in tort will be best served by pure comparison. The rule of Art. 2212a, barring recovery by a plaintiff found to be 51% or more responsible, was a legislative compromise in negligence cases, a holdover from contributory negligence. No sound reason exists for allowing a defendant to escape liability which the jury has allocated to him, irrespective of the responsibility allocated to the plaintiff. See Murray v. Fairbanks Morse, 610 F.2d 149, 162 (3d Cir.1979). Accordingly, we decline to judicially engraft the rule contained in Art. 2212a barring recovery by a plaintiff found to be 51% or more responsible. We therefore hold that in products liability cases in which at least one defendant is found liable on a theory other than negligence, the plaintiffs damages shall be reduced only by the percentage of causation attributed to the plaintiff, regardless of how large or small that percentage may be. A plaintiff may recover the percentage of damages caused by the defendants, even though his own share of causation is greater than that attributed to the defendants individually or combined.
Additionally, each defendant found to have been a cause of the plaintiff’s injuries shall be jointly and severally liable for the entire amount that the plaintiff is entitled to recover, subject to a right of contribution for payments in excess of the defendant’s percentage share. We reject any limitation on joint and several liability such as that contained in Art. 2212a § 2(c). The existence of joint and several liability becomes important when one or more defendants are insolvent. When a defendant is insolvent, the goal of allocating the loss among those responsible cannot be achieved. Nevertheless, joint and several liability in such cases furthers the fundamental policy of tort law to compensate those who are injured.9
Imposing the risk of an insolvent defendant on the remaining defendants is justified as a matter of policy because the defendants’ conduct or products endangered another person, the plaintiff, while the plaintiff’s conduct only endangered himself. See American Motorcycle Association v. Superior Court, 20 Cal.3d 578, 578 P.2d 899, 146 Cal.Rptr. 182 (1978). Furthermore, the plaintiff seeks recovery for physical injuries, but defendants’ claims for contribution are merely economic. Finally, a solvent manufacturer is better able to spread the loss than is the plaintiff.
Article 2212a will, of course, continue to govern cases in which the plaintiff alleges only negligence or where the plaintiff fails to obtain findings of defect and producing cause, or breach of warranty, against a product supplier who has been joined with a negligent defendant.
D. The Effect of Partial Settlements
We must also address the question of how a settlement with one tortfeasor affects the remaining defendants’ liability and the plaintiff’s recovery. We hold that in multiple defendant cases in which grounds of recovery other than negligence are established, the non-settling defendants’ liability and the plaintiff’s recovery shall be reduced by the percent share of causation assigned to the settling tort-feasor by the trier of fact.
The effect of a partial settlement in cases involving negligent tortfeasors was addressed by this court in Palestine *430Contractors, Inc. v. Perkins, 386 S.W.2d 764 (Tex.1964). In Palestine the court was faced with a choice between allowing a dollar credit or a pro rata credit and chose the latter. This court approved a pro rata credit because the rule allowed finality of settlement without prejudicing the non-settling defendants’ rights. A pro rata credit reduces the non-settling defendants’ liability by an amount equal to the number of settling tortfeasors divided by the total number of tortfeasors. Allowing a pro rata credit removes the incentive for collusive settlements because the plaintiff bears the risk of accepting an inadequate settlement. The court considered it fair to impose this risk on plaintiffs inasmuch as plaintiffs may avoid the risk entirely by rejecting the settlement offers. Id. at 771-73.
A dollar credit reduces the liability of the non-settling defendants, pro tanto, by the dollar amount of any settlement. The defendant’s liability thus may fluctuate depending on the amount of a settlement to which he was not a party. This fluctuation cannot be reconciled with the policy of apportioning liability in relation to each party’s responsibility, the conceptual basis of comparative causation. A dollar credit also encourages collusion by shielding plaintiffs from the effect of bad settlements while denying them the benefit of good settlements. Allowing a dollar credit does, however, ensure one recovery. Id.
Even though the reasoning in Palestine Contractors remains sound, we do not consider pro rata credit to be suitable in a comparative causation system. Pro rata allocation merely divides the damages by the number of defendants without considering their relative roles in causing the injuries. In the twenty years since Palestine, the law has advanced from crude he-adcounting to a more refined percent allocation of liability based on the relative harm caused by each defendant. Compare Art. 2212, with Art. 2212a. Accordingly, we hold that a settlement with one tort-feasor will reduce the liability of the non-settling defendants by the percentage of causation allocated to the settling tort-feasor rather than by a pro rata share. The term “tortfeasor” includes those whose liability is based on strict products liability, breach of warranty, and negligence.10
A percent credit necessarily means that settling plaintiffs may recover more than the amount of damages ultimately determined, but they also may recover less. Plaintiffs bear the risk of poor settlements; logic and equity dictate that the benefit of good settlements should also be theirs. One objection to giving plaintiffs this benefit is that doing so would violate the one recovery rule of Bradshaw v. Baylor University, 126 Tex. 99, 84 S.W.2d 703 (1935). This case thus requires that we answer a question expressly left open in Cypress Creek v. Muller, 640 S.W.2d 860 (Tex.1982) —how does the Bradshaw one recovery rule apply in cases not controlled by Art. 2212a?
In stating the one recovery rule, the court in Bradshaw relied on Hunt v. Ziegler, 271 S.W. 936 (Tex.Civ.App.—San Antonio 1925), aff'd 280 S.W. 546 (Tex. Comm’n App.1926, judgmt. adopted). See T.L. James & Co. v. Statham, 558 S.W.2d 865 (Tex.1977). The rationale in Hunt v. *431Ziegler indicates that the one recovery rule was originally adopted because the courts could not conceive of allocating liability. Injuries were considered indivisible; therefore, a settling defendant could only offer to pay for the whole injury, not just his part. A settlement greater than or equal to the amount of damages ultimately found by the trier of fact left nothing to pay for. Consequently, the plaintiff had no remaining injury for which he could recover.
The reasoning behind the one recovery rule no longer applies. The system of comparative causation we adopt allows allocation of liability between the parties, even when the injury itself is indivisible. Cf Art. 2212a (apportioning liability in negligence cases). Because each defendant’s share can now be determined, it logically follows that each may settle just that portion of the plaintiffs suit. The settlement does not affect the amount of harm caused by the remaining defendants and likewise should not affect their liability.
An additional rationale supporting the one recovery rule was that it prevented unjust enrichment; however, any enrichment of plaintiffs under the new system of comparative causation is not unjust, for the simple reason that no one is harmed. The settling defendant cannot complain, because he agreed to pay. The non-settling defendant has no right to complain, because he was not a party to and is not affected by the settlement. Article 2212a § 2(e) provides for the same result, and we rejected the unjust enrichment argument when Art. 2212a § 2(e) applies, saying: Cypress Creek Utility Service Co. v. Muller, 640 S.W.2d at 866-67 (Tex.1982) (quoting Theobald v. Angelos, 44 N.J. 228, 208 A.2d 129 (1965)); see also Dobson v. Camden, 705 F.2d at 769; Fleming, Report to the Joint Committee of the California Legislature on Tort Liability on the Problems Associated with American Motorcycle Association v. Superior Court, 30 Hastings L.J. 1465, 1497-98; Comment, Comparative Negligence, Multiple Parties, and Settlements, 65 Calif.L.Rev. 1264, 1278-79 (1977). The unjust enrichment argument is equally without merit when Art. 2212a does not apply.
If there is enrichment, it is not at the defendant’s expense. Defendant does not seek to make [the settling defendant] whole but rather to profit from the injustice that [the settling defendant] supposedly experienced. If the voluntary agreement between plaintiff and [the settling defendant] were thought so to offend public policy as to require redress, the remedy would run to [the settling defendant] rather than to a stranger to the bargain.
This court stated in dictum in McMillen v. Klingensmith, 467 S.W.2d 193 (Tex.1971) that “a claimant in no event will be entitled to recover more than the amount required for full satisfaction of his damages.” Although that statement on its face is inconsistent with the system we here adopt, the reasoning in Klingensmith actually supports our adoption of percent credit. Klingensmith rejected the unity of release rule partly to avoid rewarding tort-feasors who were unwilling to settle. Id. at 196. Non-settling tortfeasors will likewise profit if they are allowed to benefit from a generous settlement in which they refused to participate. This unfair result would follow from strict application of the one recovery rule and should not be approved. A percent credit system requires the non-settling defendant to pay his allocated share, neither enhanced nor diminished by the accommodation between the plaintiff and a settling tortfeasor.
Finally, settlement dollars are not the same as damages. Settlement dollars represent a contractual estimate of the value of the settling tortfeasor’s liability and may be more or less than the proportionate share of the plaintiffs damages. The settlement includes not only damages, but also the value of avoiding the risk, expense, and adverse public exposure that accompany going to trial. There is no conceptual inconsistency in allowing a plaintiff to recover more from a settlement or partial settlement than he could receive as dam*432ages. Leger v. Drilling Well Control, 592 F.2d at 1250 & n. 10; Fleming, at 1497.
For the reasons stated, we hold that the one recovery rule does not prevent our adopting a system that reduces the plaintiffs recovery and the non-settling defendants’ liability by the percentage of causation assigned to any tortfeasor with whom plaintiff has settled. We hereby adopt a percent credit rule, which will leave defendants unaffected by settlements in which they do not participate. Each party’s share of liability will be that determined by the jury. Plaintiffs will benefit from good settlements and bear the risk of bad ones, just as they do in single-tortfeasor cases. Allowing plaintiffs to keep the excess from a good settlements may violate the one recovery rule, but no one is harmed. Accordingly, to the extent it conflicts with this opinion, we overrule Bradshaw v. Baylor University.
E. Remaining Issues
Because of the scope of this opinion, we consider it appropriate to note two important rules which remain unchanged. We reaffirm the rule in § 402A comment n of the Restatement (Second) of Torts that negligent failure to discover or guard against a product defect is not a defense. Shamrock Fuel & Oil Sales Co. v. Tunks, 416 S.W.2d 779 (Tex.1967); McKisson v. Sales Affiliates, Inc., 416 S.W.2d 787 (Tex.1967). This rule serves to protect consumer reliance on product safety. E.g., Shamrock, 416 S.W.2d at 786; West v. Caterpillar Tractor Co., 336 So.2d 80, 92 (Fla.1976); Busch v. Busch Construction, Inc., 262 N.W.2d 377, 394 (Minn.1977); Star Furniture Co. v. Pulaski Furniture Co., 297 S.E.2d 854, 862-63 (W.Ya.1982).
Comparative causation does not affect the right of a retailer or other member of the marketing chain to receive indemnity from the manufacturer of the defective product when the retailer or other member of the marketing chain is merely a conduit for the defective product and is not independently culpable. See General Motors Corp. v. Simmons, 558 S.W.2d 855, 860-61 (Tex.1977); see also Edgar, supra at 47. Compare Thiele v. Chick, 631 S.W.2d 526 (Tex.App.—Houston [1st Dist.] 1982, writ ref’d n.r.e.) (seller allowed indemnity where no independent culpability shown); with United Tractor, Inc. v. Chrysler Corp., 563 S.W.2d 850 (Tex.Civ.App.—El Paso 1978, no writ) (independently negligent manufacturer-assembler denied indemnity); and Ford Motor Co. v. Russell & Smith Ford Co., 474 S.W.2d 549 (Tex.Civ.App—Houston [14th Dist.] 1971, no writ) (independently negligent retailer denied indemnity). If there is no allegation of independent liability, the liability of the members of the marketing chain should be submitted as one. The jury will determine the causation attributable to the defective product. Of course, each defendant member of the marketing chain remains liable to the plaintiff. See Lubbock Manufacturing Co. v. International Harvester Co., 584 S.W.2d 908 (Tex.Civ.App.—Dallas 1979, writ ref’d n.r.e.).
F. Conclusion
Having adopted comparative causation for strict products liability actions, we turn to its application in this case. Cessna asks us to reverse the judgments below and to remand for a new trial allowing for an apportionment of liability. Duncan argues that Cessna waived any rights to contribution or indemnity based on the alleged negligence of Parker or Smithson.
The answer depends on whether Cessna had live pleadings at trial raising the issue of negligence in support of its claims for contribution and indemnity. If there were pleadings, it was encumbent upon Cessna to offer proof, and if its offer of proof were rejected, to make out a bill of exceptions showing the substance of the excluded evidence. Gulf Paving Co. v. Lofstedt, 144 Tex. 17, 188 S.W.2d 155, 159 (1945); J. Weingarten, Inc. v. Brockman, 134 Tex. 451, 135 S.W.2d 698, 699 (1940); Biggins v. Gulf, C. & S.F. Ry. Co., 102 Tex. 417, 118 S.W. 125, 126 (1909); Harris v. Harris, 605 S.W.2d 684, 686-87 (Tex.Civ.App.—Houston [1st Dist.] 1980, writ ref’d n.r.e.); Don*433ald R. Reedy & Co. v. Jenkins & Co., 597 S.W.2d 62, 63 (Tex.Civ.App.—Dallas 1980, writ ref d n.r.e.). On the other hand, if all of Cessna’s pleadings alleging negligence were struck, Cessna had no duty to offer testimony or to perfect a bill of exceptions. Minus v. Doyle, 141 Tex. 67, 170 S.W.2d 220, 223 (1943).
Cessna filed a counterclaim on May 9, 1979 alleging negligence as the basis for a right of contribution or indemnity. On May 14, 1979 Duncan filed a motion to strike. Before the trial court acted on Duncan’s motion, Cessna filed its first amended original answer on January 11, 1980, also alleging Duncan’s negligence. The trial court struck Cessna’s counterclaim on January 18, 1980, and in response to Duncan’s special exceptions and motion to strike, the trial court also struck Cessna’s first amended original answer by order of April 30, 1980.
The dissent stops at this point and concludes that the trial court “sustained plaintiff’s exceptions to all of Cessna’s negligence pleadings.” The record, however, contradicts this statement. After the trial court struck Cessna’s counterclaim, but before any action was taken on the first amended original answer, Cessna filed its second amended answer on March 23,1980. In the 'second answer, Cessna reasserted negligence, despite the trial court’s January 18th order. No special exception or motion to strike was ever leveled against this pleading, nor was any order striking the second answer ever entered. As a consequence, Cessna’s pleadings contained live negligence allegations requiring proof.11
Cessna offered deposition testimony to prove the alleged negligence of Duncan, but the trial court refused to admit such evidence. At that point, Cessna did not even attempt to perfect any proper bill of exceptions containing the excluded evidence as required by Rule 372. Without a bill of exceptions showing what the excluded testimony would have been, we cannot determine whether the testimony was inadmissible on some other ground or whether its exclusion was otherwise harmless error. Under the authorities cited above, a finding of reversible error is therefore unwarranted.
Cessna also maintains that we should remand in the interest of justice as provided by Rule 505. In effect, we are asked to overlook Cessna’s reassertion of negligence in disregard of the trial court’s order. See Sandstrum v. Magruder, 510 S.W.2d 388 (Tex.Civ.App.—Houston [1st Dist.] 1974, writ ref’d n.r.e.); Ship Ahoy, Inc. v. Whalen, 347 S.W.2d 662 (Tex.Civ.App.—Houston 1961, no writ); Shaw v. Universal Life & Accident Insurance Co., 123 S.W.2d 738 (Tex.Civ.App.—Dallas 1938, writ dism’d judgmt cor.); see also Gehrke v. State, 363 S.W.2d 490 (Tex.Civ.App.—San Antonio 1962, writ ref’d n.r.e.); City of Alice v. Lacey, 362 S.W.2d 919 (Tex.Civ.App.—San Antonio 1962, no writ). Whatever Cessna hoped to gain by defying the trial court’s prior rejection of identical allegations of negligence, it cannot now expect to avoid the legal effect of its pleadings.
Our holding that a remand in this case will not serve the interests of justice does not conflict with any prior decision of this court. The eases relied on by Cessna and the dissent fall within two distinct categories. In the first category, the party seeking a remand had presented evidence that might allow relief under the proper theory on remand. Without a bill of exceptions, the record before us contains no such evidence. In the second category, a remand was justified because the new theory or rule required new and different evidence.
Under the existing rule of Palestine Contractors, Inc. v. Perkins, 386 S.W.2d 764 (Tex.1964), Cessna tried to prove that the settling parties were liable to Duncan in order to get a pro rata reduc*434tion of its liability. Cessna offered evidence of Smithson’s negligence for this purpose, but failed to preserve its assignment of error when the evidence was excluded. If this court were to remand for a new trial, giving Cessna the benefit of comparative causation, Cessna would be required to introduce exactly the same type of evidence. Allowing Cessna to offer the same proof on retrial that it failed to preserve by bill of exceptions would emasculate the elementary rule that an assignment of error relating to the exclusion of evidence must be preserved in a proper bill of exceptions. E.g., Gulf Paving Co. v. Lofstedt, 144 Tex. 17, 188 S.W.2d 155 (1945). Cessna is not being penalized for failing to anticipate our adoption of comparative causation. Rather, Cessna’s loss stems from its failure to comply with the law as it existed.12
In its motion for rehearing, Cessna alternatively requests that its liability to Duncan be reduced by the dollar amount of Duncan’s prior settlement. Decisions such as Leong v. Wright, 478 S.W.2d 839 (Tex.Civ.App.—Houston [14th Dist.] 1972, writ ref’d n.r.e.) and Petco Corp. v. Plummer, 392 S.W.2d 163 (Tex.Civ.App—Dallas 1965, writ ref’d n.r.e.) support this request. Both lower courts allowed a dollar credit rather than a pro rata reduction because the liability of the settling parties was not established. These cases reflect the accepted understanding of the effect of the one recovery rule of Bradshaw v. Baylor University. Although we have disapproved allowing dollar credits in future cases in which comparative causation applies, Cessna is entitled to a credit in this case. Accordingly, Cessna’s liability shall be reduced by the amount of the prior settlement agreement.
In future cases, however, when Art. 2212a does not apply, the basis of any right of contribution will be joint and several liability. If the settling tortfeasor’s liability is not established, the defendant will not be entitled to any benefit from the settlement agreement. For the reasons stated, the rationale behind the one recovery rule in Bradshaw does not require a dollar credit, and we see no other reason to allow such a reduction.13
Finally, we must determine whether comparative causation for strict liability in tort should apply prospectively or retrospectively. This matter is within our discretion. Sanchez v. Schindler, 651 S.W.2d 249, 254 (Tex.1983). Although our decisions usually apply retrospectively, “exceptions are recognized when considerations of fairness and policy” dictate prospective effect only. Id. It could be unfair to apply this decision retrospectively, because litigants and trial courts have justifiably relied on our discussion of articles 2212 and 2212a in Simmons for six years. Accordingly, comparative causation will govern all products cases tried after July 13,1983, the date of our original opinion, including all cases remanded for retrial for whatever reason.
The judgments of the trial court and court of appeals are reversed, and judgment is here rendered on the jury verdict for Duncan, subject to a credit for the $90,000 already received in settlement.
POPE, C.J., concurs and dissents in an opinion in which McGEE, BARROW and CAMPBELL, JJ., join.. Duncan remarried after Parker’s death.
. These actions were tried together, but have been separated on appeal to this court. Smithson’s case presents other legal issues. See Smithson v. Cessna Aircraft Co., 665 S.W.2d 439 (Tex.1984).
. Although a federal court in United States v. Reilly, 385 F.2d 225 (10th Cir.1967), applying New Mexico law reached a contrary conclusion — that a general release was ineffective to discharge an unnamed tortfeasor — our discussion assumes that the state court in Johnson states New Mexico law.
. Cessna also argues that § 170(1) of the Second Restatement applies. Section 170(1) provides,
The law selected by application of the rule of § 145 [governing tort issues] determines the effect of a release ... given to one joint tort-feasor upon the liability of the other.
Because Cessna was not a party to the Duncan release, however, its liability to Duncan implicates policies underlying both contract and tort law. See generally R. Leflar, American Conflicts Law § 134, at 273 (3d ed. 1977). We therefore decline to adopt § 170(1).
.Section 6, stating the general principles, provides,
*421(1) A court, subject to constitutional restrictions, will follow a statutory directive of its own state on choice of law.
(2) When there is no such directive, the factors relevant to the choice of the applicable rule of law include
(a) the needs of the interstate and international systems,
(b) the relevant policies of the forum,
(c) the relevant policies of other interested states and the relative interests of those states in the determination of the particular issue,
(d) the protection of justified expectations,
(e) the basic policies underlying the particular field of law,
(f) certainty, predictability and uniformity of result, and
(g) ease in the determination and application of the law to be applied.
. As for the application of Kansas law, since *422Cessna has not asserted any error in the trial court’s application of New Mexico law, we need not decide whether the resolution of this issue will impair any policies expressed in Kansas law.
. Furthermore, most of those courts that have rejected such a system for loss allocation did not do so on policy grounds; rather, they believed that state comparative negligence statutes did not allow it. See, e.g., Melia v. Ford Motor Co., 534 F.2d 795 (8th Cir.1976) (applying Nebraska law); Kirkland v. General Motors Corp., 521 P.2d 1353 (Okla.1974); cf. Kinard v. Coats, 37 Colo.App. 555, 553 ,P.2d 835 (1976) (holding that statute did not apply because strict products liability is premised on the policy of enterprise liability, not on negligence principles).
. We suggest that the following general form, adapted to the specific allegations of each case, provides a general idea of the submission of the comparative apportionment issue in these cases:
If, in answer to Questions —, —, and —, you have found that more than one party’s act(s) or product(s) contributed to cause the plaintiffs injuries, and only in that event, then answer the following question.
Find from a preponderance of the evidence the percentage of plaintiff’s injuries caused by:
Product X _
Defendant Y _
Plaintiff Z _
Total 100%
Cf. Daly v. General Motors Corp., 20 Cal.3d 725, 575 P.2d 1162, 144 Cal.Rptr. 380 (1978) *428(suggesting analogous special issues from maritime seaworthiness cases).
. An alternative would be to reallocate the insolvent tortfeasor’s share of liability among all parties whose actions or products were a cause of the injuries, including the negligent plaintiff. This suggestion is attractive and was endorsed by a distinguished Special Committee of the Tort and Compensation Section of the State Bar. As a judicial rule, however, reallocating the insolvent’s share would create problems of post-trial jurisdiction and finality of judgments. See TEX.R.CIV.P. 329b.
. Even if we disregarded the opinion in Palestine Contractors, we would favor a percent credit instead of a dollar credit. This approach is taken in the new Uniform Comparative Fault Act § 6 (1977), which supersedes the dollar credit of the 1955 Uniform Contribution Among Tortfeasors Act § 4. The old dollar credit is considered inappropriate in jurisdictions adopting a system of comparative causation. Proportionate reduction of the non-settling defendant’s liability has also received approval from other courts and commentators. See, e.g., Dobson v. Camden, 705 F.2d 759 (5th Cir.1983); Leger v. Drilling Well Control, Inc., 592 F.2d 1246 (5th Cir.1979); River Garden Farms, Inc. v. Superior Court, 26 Cal.App.3d 986, 103 Cal.Rptr. 498 (1972); Fleming, Report to the Joint Committee of the California Legislature on Tort Liability on the Problems Associated with American Motorcycle Association v. Superior Court, 30 Hastings LJ. 1465 (1979); Special Project, Texas Tort Law in Transition, 57 Texas L.Rev. 381 (1979); Comment, Comparative Negligence, Multiple Parties, and Settlements, 65 Calif.L.Rev. 1264 (1977); Comment, Settlement in Joint Tort Cases, 18 Stan.L.Rev. 486 (1966).
. We note that the dissent meticulously identifies the date each of Cessna’s other pleadings were struck yet fails to identify any pleading or order attacking Cessna’s second amended original answer. The excerpts from the record relied on by the dissent establish only that Cessna's second answer probably would have been struck, not that it was.
. The dissent argues that because we gave the winning party in Sanchez v. Schindler, 651 S.W.2d 249 (Tex.1983), the benefit of the rule we adopted in that case, we should also remand this case so that Cessna may benefit from our adoption of comparative causation. In Sanchez, however, the party successfully urging the change in the law preserved her error at each stage of the litigation; Cessna failed to preserve error in the trial of this case, thereby losing its right to a new trial.
. We do not agree that if all parties consent to a dollar credit, the case may be tried more efficiently because the jury will not have to consider the settling defendant’s liability. See Special Project, Texas Tort Law in Transition, 57 Texas L.Rev. 381, 453 (1979). Trials would be artificial rather than efficient if jurors were required to allocate causation without considering all of the causes. Furthermore, any benefit would only arise in the unlikely event that the parties agreed to dollar credit before trial.