I
Repudiating the existing contributory negligence system and adopting a system of comparative negligence, this court in Li v. Yellow Cab Co. (1975) 13 Cal.3d 804 [119 Cal.Rptr. 858, 532 P.2d 1226, 78 A.L.R.3d 393] repeatedly—like the tolling bell—enunciated the principle that the extent of liability must be governed by the extent of fault. Thus, the court stated, “the extent of fault should govern the extent of liability” (id., at p. 811), “liability for damage will be borne by those whose negligence caused it in direct proportion to their respective fault” (id., at p. 813), and “the fundamental purpose of [the rule of pure comparative negligence] shall be to assign responsibility and liability for damage in direct proportion to the amount of negligence of each of the parties” (id., at p. 829). And in a cacophony of emphasis this court explained that the “basic objection to the doctrine [of contributory negligence]—grounded in the primal concept that in a system in which liability is based on fault, the extent of fault should govern the extent of liability—remains irresistible to reason and all intelligent notions of fairness.” (Id., at p. 811.)
*609Now, only three years later, the majority of my colleagues conclude that the Li principle is not irresistible after all. Today, in the first decision of this court since Li explaining the operation of the Li principle, they reject it for almost all cases involving multiple parties.
The majority reject the Li principle in two ways. First, they reject it by adopting joint and several liability holding that each defendant—including the marginally negligent one—will be responsible for the loss attributable to his codefendant’s negligence. To illustrate, if we assume that the plaintiff is found 30 percent at fault, the first defendant 60 percent, and a second defendant 10 percent, the plaintiff under the majority’s decision is entitled to a judgment for 70 percent of the loss against each defendant, and the defendant found only 10 percent at fault may have to pay 70 percent of the loss if his codefendant is unable to respond in damages.
The second way in which the majority reject Li’s irresistible principle is by its settlement rules. Under the majority opinion, a good faith settlement releases the settling tortfeasor from further liability, and the “plaintiff’s recovery from nonsettling tortfeasors should be diminished only by the amount that the plaintiff has actually recovered in a good faith settlement, rather than by an amount measured by the settling tortfeasor’s proportionate responsibility for the injury.” (Ante, p. 604.)1 The settlement rules announced today may turn Li’s principle upside down—the extent of dollar liability may end up in inverse relation to fault.
Whereas the joint and several liability rules violate the Li principle when one or more defendants are absent or unable to respond in damages, the settlement rules will ordinarily preclude effecting the majority’s principle in cases when all defendants are involved in the *610litigation and are solvent. To return to my 30-60-10 illustration and further assuming both defendants are solvent, the plaintiff is ordinarily eager to settle quickly to avoid the long delay incident to trial. Further, he will be willing to settle with either defendant because under the majority’s suggested rules, he may then pursue the remaining defendant for the balance of the recoverable loss (70 percent) irrespective whether the remaining defendant was 10 percent at fault or 60 percent at fault. The defendants’ settlement postures will differ substantially. Realizing the plaintiff is eager for quick recovery and is capable of pursuing the codefendant, the defendant 60 percent liable for the loss will be prompted to offer a sum substantially below his share of fault, probably paying 20 to 40 percent of the loss. The defendant only 10 percent at fault will be opposed to such settlement, wishing to limit his liability. To compete with his codefendant in settlement offers he will be required to offer substantially in excess of his 10 percent share of the loss, again frustrating the Li principle that the extent of liability should be governed by the extent of fault. Should he fail to settle, the 10 percent at fault defendant runs the risk that his codefendant will settle early for perhaps half of his own liability, while the lesser negligent person must eventually pay the remainder, not only frustrating the Li principle but turning it upside down. In any event, it is extremely unlikely he can settle for his 10 percent share.2
*611The foregoing demonstrates that under the majority’s joint and several liability and settlement rules, only rarely will the Li principle be carried out in multi-party litigation. The principle will be frustrated if one or more defendants are unavailable, insolvent, or have settled. Prior to Li, the overwhelming majority of accident cases were settled in whole or in part, and assuming this practice continues, the Li principle will not be realized in those cases. In a substantial number of the remaining cases it can be expected that one of the tortfeasors will not be able to respond in damages, again frustrating the Li principle. In sum, although the majority devote approximately half of their opinion to asserted maintenance of the Li principle (pts. 3, 4, and 5), in only a very small number of multiple party cases will the loss be shared in accordance with that principle.
Attempting to justify their repudiation of the Li principle in favor of joint and several liability, the majority suggest three rationales. First, we are told that the feasibility of apportioning fault on a comparative basis does not “render an indivisible injury ‘divisible,’ ” each defendant’s negligence remaining a proximate cause of the entire indivisible injury. (Ante, p. 588.) The argument proves too much. Plaintiff negligence is also a proximate cause of the entire indivisible injury, and the argument, if meritorious, would warrant repudiation of Li not only in the multiple party case but in all cases.
The second rationale of the majority lies in two parts. First, we are told that after Li there is no reason to assume that plaintiffs will “invariably” be guilty of negligence. (Ante, p. 589.) Obviously this is true. The basis of joint and several liability prior to Li was that between an innocent plaintiff and two or more negligent defendants, it was proper to hold the defendants jointly and severally liable. The innocent plaintiff should not suffer as against a wrongdoing defendant. (Ante, p. 588.) (Finnegan v. Royal Realty Co. (1950) 35 Cal.2d 409, 433-434 [218 P.2d 17].) Accordingly, it is not unreasonable to reject the Li principle when we are comparing the plaintiff’s innocence and defendants’ negligence. But the issue presented by this case is whether joint and several liability shall be extended to Li cases, cases where the plaintiff by definition is negligent. While we cannot know whether a plaintiff will be found negligent until trial, we also cannot know whether any given defendant will be found at fault until trial. Since liability is not to be determined until after trial, there is no reason not to deal with the real issue before us whether joint and several liability should be applied in cases where the plaintiff is *612found negligent—i.e., cases where by definition the plaintiff is “invariably” found negligent.
As a second part of the second rationale for joint and several liability we are told that a plaintiff’s culpability is not equivalent to that of a defendant. This is obviously true—this is what Li is all about. The plaintiff may have been driving 50 miles in excess of the speed limit while the defendants may have been driving 10 miles in excess. The converse may also be true. But the differences warrant departure from the Li principle in toto or not at all.
The majority’s third rationale for rejecting the Li principle is an asserted public policy for fully compensating accident victims. The majority state that joint and several liability “recognizes that fairness dictates that the ‘wronged party should not be deprived of his right to redress,’ but that ‘[t]he wrongdoers should be left to work out between themselves any apportionment.’ (Summers v. Tice (1948) 33 Cal.2d 80, 88 [199 P.2d 1, 5 A.L.R.2d 91].)” (Ante, p. 590.) The quoted language.is not helpful to the majority when the plaintiff is also negligent because he is himself a wrongdoer.
Until today neither policy nor law called for fully compensating the negligent plaintiff. Prior to Li, the negligent plaintiff was denied all recovery under the contributory negligence doctrine—the policy reflected being directly contrary to that asserted today. Li, of course, repudiated that doctrine replacing it with a policy permitting compensation of the negligent accident victim but only on the basis of comparative fault. Moreover, Li cannot be twisted to establish a public policy requiring rejection of its own irresistible principle. In sum, the majority are establishing a new policy both contrary to that existing prior to Li and going further than that reflected by the comparative principle enunciated in Li.
Conceivably, such a new public policy departing from intelligent notions of fairness may be warranted but, if so, its establishment should be left for the Legislature. Before going beyond Li’s principle “irresistible to reason and all intelligent notions of fairness” (13 Cal.3d at p. 811), a full evaluation should be made of society’s compensation to accident victims through our tort system in comparison to all other means used by society to compensate victims. A study should include such matters as the relative workings of the liability insurance system in providing benefits, disability insurance and employer benefits, medical insurance, *613workers’ compensation, insurance against uninsured defendants, Medicare, Medi-Cal and the welfare system. Reconsideration of the Collateral source rule would also be required before adoption of a public policy going beyond intelligent notions of fairness. The evidence gathering and hearings necessary for the requisite study are within the capabilities of the Legislature; this court is institutionally incapable of undertaking it.
The majority rely on decisions from Mississippi, New York, Wisconsin, and Georgia for the proposition that courts have retained joint and several liability under comparative negligence. (Ante, p. 591.) In the cases cited from the first two jurisdictions, it does not appear that the plaintiff was negligent under the facts or that the court in adhering to joint and several liability was considering cases where the plaintiff was negligent. Thus, those cases stand for nothing more than application of joint and several liability when a plaintiff is innocent and the defendants are guilty, the traditional common law application. The third jurisdiction, Wisconsin, is not a pure comparative negligence jurisdiction. Rather, the negligent plaintiff can recover only if his “ ‘negligence was not as great as the negligence of the person against whom recovery is sought.’ ” (Chille v. Howell (1967) 34 Wis.2d 491 [149 N.W.2d 600, 604].) Because of the limitation on recovery by negligent plaintiffs in Wisconsin, it may be justifiable to apply joint and several liability by analogy to the common law principle that as between an innocent plaintiff and any negligent defendant, the entire loss shall fall on the negligent actor. Obviously, such justification is not available in a pure comparative jurisdiction like California. Only the Georgia case is in point.
In any event as pointed out by Justice Thompson in the opinion and chart prepared in the Court of Appeal in this case, several jurisdictions adopting comparative fault have abolished joint and several ¿ability.3
In my view the majority’s effort to resist the irresistible fails. They have furnished no substantial reason for refusing to apply the Li principle to multi-party litigation.
II
Adherence to the Li principle that the extent of liability is governed by the extent of fault requires that only á limited form of joint and several *614liability be retained in cases where the plaintiff is negligent.4 The issue of joint and several liability presents the problem whether the plaintiff or the solvent defendants should bear the portion of the loss attributable to unknown defendants or defendants who will not respond in damages due to lack of funds.
Consistent with the Li principle—the extent of liability is governed by the extent of fault—the loss attributable to the inability of one defendant to respond in damages should be apportioned between the negligent plaintiff and the solvent negligent defendant in relation to their fault. (Fleming, Foreword: Comparative Negligence At Last—By Judicial Choice (1976) 64 Cal.L.Rev. 239, 251-252, 257-258.) Returning to my 30-60-10 illustration, if the 60 percent at fault defendant is unable to respond, the 30 percent at fault plaintiff should be permitted to recover 25 percent of the entire loss from the 10 percent at fault solvent defendant based on the 3 to 1 ratio of fault between them. (The solvent defendant would have added to his 10 percent liability one-fourth of the 60 percent or 15 percent to reach the 25 percent figure.) To the extent that anything is recovered from the 60 percent at fault defendant, the money should be apportioned on the basis of the 3 to 1 ratio. The system is based on simple mechanical calculations from the jury findings.
Placing^ the entire loss attributable to the insolvent defendant solely on the negligent plaintiff or solely on the solvent negligent defendant is not only contrary to the Li principle, but also undermines the entire system of comparative fault. If the portion attributable to the insolvent defendant is placed upon the negligent plaintiff, the solvent defendant will attempt to reduce his liability by magnifying the fault of the insolvent defendant. Should the insolvent’s portion be placed solely upon the solvent defendant—as done by the majority’s application of joint and several liability—the plaintiff will have an incentive to magnify the fault of the insolvent defendant.5 Because the insolvent—and *615therefore disinterested—defendant will usually not be present at trial to defend himself, any semblance to comparative fault will be destroyed.
Similarly, settlement rules should also reflect the Li principle. When a defendant settles, he should be deemed to have settled his share of the total liability and the pleadings and releases should so reflect. The nonsettling defendant should be liable only for the portion of the loss attributable to him—deducting from the total loss the amount attributable to the plaintiff’s negligence6 and the amount attributable to the settling defendant’s negligence. This rule adopted by Wisconsin (Pierringer v. Hoger (1963) 21 Wis.2d 182 [124 N.W.2d 106, 111-112]), would force a plaintiff to demand settlements reasonably commensurate to the fault of the settling defendant because he will no longer be able to settle quickly and cheaply, then holding the remaining defendants for part of his codefendant’s share of the loss. Granted, the nonsettling defendant will have an incentive to magnify the fault of the settling defendant, but it is not unfair to place the burden of defending the settling defendant upon the plaintiff for three reasons: He is the one who chose to settle, the settlement has eliminated any right of contribution or partial indemnity of the nonsettling defendant, and the plaintiff in obtaining his settlement may secure the cooperation of the settling defendant for the later trial.
Ill
“[irresistible to reason and all intelligent notions of fairness” (13 Cal.3d 804, 811), this court created a policy three years ago the majority today cavalierly reject without real explanation. Their attempted rationale for rejection of the Li principle insofar as it is based on a newly discovered public policy is entitled to little weight. The public has no such policy and any attack on the principle based on logic or abstract notions of fairness fail. The principle is transparently irresistible in the abstract.
If not applied across the board the Li principle should be abandoned. The reason for abandonment applies not only to multi-party cases but also to two-party cases, warranting total repudiation of the principle, not merely the majority’s partial rejection.
*616While logically reasonable and fair in the abstract, the Li principle is generally unworkable, producing unpredictable and inconsistent results. Implementation of the principle requires judgment beyond the ability of human judges and juries. The point is easily illustrated. If the first party to an accident drove 10 miles in excess of the speed limit, the second 50 miles in excess, it is clear that the second should suffer the lion’s share of the loss. But should he pay 55 percent of the loss, 95 percent or something in between? That question cannot be answered with any precision, and human beings will not answer it consistently. Yet that is the easiest question presented in comparing fault because we are dealing only with apples. When we add oranges to the comparison, there are no guidelines. If the first driver also was driving under the influence of Jack Daniels, reasonable judges and juries will disagree as to who shall bear the lion’s share of the loss, much less the percentages. Finally, when the case is pure apples and oranges—one party speeds, the other runs a stop signal—there is no guide post, much less guidelines, and acting in furtherance of the Li principle, reasonable judges and juries can be expected to come up with radically different evaluations.7
In short, the pure comparative fault system adopted by Li not only invites but demands arbitrary determinations by judges and juries, turning them free to allocate the loss as their sympathies direct. We may expect that allocation of the. loss will be based upon the parties’ appearance and personality and the abilities of their respective counsel. The system is a nonlaw system. Furthermore, prior to Li our tort system of liability was condemned because it was so inefficient in transferring the liability insurance premium to the accident victim (e.g., Conard et al., Automobile Accident Costs and Payments (1964) pp. 58-61). The complexities and unpredictability of the Li system can only make the system even more inefficient.
I do not suggest return to the old contributoiy negligence system. The true criticism of that system remains valid: one party should not be required to bear a loss which by definition two have caused. However, in departing from the old system of contributory negligence numerous approaches are open, but the Legislature rather than this court is the *617proper institution in a democratic society to choose the course. To accommodate the true criticism, for example, it might be proper to take the position that a negligent plaintiff forfeits part—but not all—of his recovery in a percentage fixed by the Legislature. A fixed percentage approach would eliminate the impossible task of comparing apples and oranges placed upon the trier of fact by Li and would provide the consistency, certainty and predictability which foster compromise and settlement. Although the percentage would be arbitrary, the allocation of loss as demonstrated above is necessarily arbitrary under the present system.
In my dissenting opinion in Li I pointed out: “[T]he Legislature is the branch best able to effect transition from contributory to comparative or some other doctrine of negligence. Numerous and differing negligence systems have been urged over the years, yet there remains widespread disagreement among both the commentators and the states as to which one is best. (See Schwartz, Comparative Negligence (1974) Appen. A, pp. 367-369 and § 21.3, fn. 40, pp. 341-342, and authorities cited therein.) This court is not an investigatory body, and we lack the means of fairly appraising the merits of these competing systems. Constrained by settled rules of judicial review, we must consider only matters within the record or susceptible to judicial notice. That this court is inadequate to the task of carefully selecting the best replacement system is reflected in the majority’s summary manner of eliminating from consideration all but two of the many competing proposals—including models adopted by some of our sister states.” (Fn. omitted; 13 Cal.3d at pp. 833-834.)
Again, it must be urged that this is a subject to which the Legislature should address itself. Not only are there a number of different approaches to plaintiff negligence in our sister states but recent years have spawned numerous studies of the problem from the societal point of view. (E.g., Cal. Citizens Com. on Tort Reform, Righting the Liability Balance (Sept. 1977).) The two most modem trends of compensating accident victims run in directly contrary approaches—the nonfault approach where negligence may be ignored and the comparative fault approach where the quantum of negligence is to be meticulously divided among the parties. No area of the law calls out more for a clear policy established by democratically elected representatives.
Petitioner’s application for a rehearing was denied March 16, 1978, and the opinions were modified to read as printed above. Clark, J., was of the opinion that the application should be granted.
Although one of the most important matters determined by today’s decision, the issue of pro rata reduction or dollar amount reduction was barely mentioned and the relative merits of the two systems were not briefed or argued by the parties or by any of the numerous amici. The overwhelming weight of authority—contrary to the majority—is for pro rata reduction rather than settlement amount reduction. (Ark. Stats. Anri., § 34-1005; Hawaii Rev.Laws § 663-15; Nebben v. Kosmalski (1976) 307 Minn. 211 [239 N.W.2d 234, 236]; Theobald v. Angeles (1965) 44 N.J. 228 [208 A.2d 129, 131]; Rogers v. Spady (1977) 147 N.J.Super. 274 [371 A.2d 285, 287]; N.Y.Gen.Obl.Law, § 15-108; R.I. Gen.Laws (1956) § 10-6-8; S.D.Codified Laws 15-8-18; Tex.Rev.Civ.Stat., art. 2212a, § 2(e); Utah Code 78-27-43; Gomes v. Brodhurst (3d Cir. 1967) 394 F.2d 465; Pierringer v. Hoger( 1963) 21 Wis.2d 182 [124 N.W.2d 106]; Wyo.Stat. Ann. § 1-7.6; but cf. Fla.Stat.Ann. § 768.31; Mass.Laws Ann., ch. 231B, § 4.) Although I believe it is improper for the court to reach such an important issue without the aid of counsel, I am compelled to discuss the problem because the majority has determined it.
In addition, the policy in favor of settlement will be frustrated by the majority’s rule that the plaintiff’s recovery against nonsettling tortfeasors should be diminished only by the amount recovered in a good faith settlement rather than by settling tortfeasor’s proportionate responsibility. (Ante, p. 604.) As the majority recognize: “ ‘Few things would be better calculated to frustrate [section 877’s] policy, and to discourage settlement of disputed tort claims, than knowledge that such a settlement lacked finality and would lead to further litigation with one’s joint tortfeasors, and perhaps further liability.’ ” (Id.) Settlement by one tortfeasor is not going to compel the other tortfeasor to withdraw his cross-complaint for total or partial indemnity. Rather there will be a claim of bad faith because if the jury awards the plaintiff all of the damages sought and concludes that the settling tortfeasor should bear the lion’s share of the responsibility for the laws, the settling tortfeasor would have escaped for a small fraction of his actual liability. This alone, although not determinative, would indicate bad faith. (River Garden Farms, Inc. v. Superior Court (1973) 26 Cal.App.3d 986, 997 [103 Cal.Rptr. 498] (“price is the immediate signal for the inquiry into good faith”).)
Obviously, in most cases the jury will not award plaintiff all of the damages sought and will not conclude the settling tortfeasor should have borne the lion’s share. But because prior to trial these matters are necessarily uncertain and the possibility of establishing bad faith exists, the nonsettling tortfeasor’s counsel must continue to maintain his cross-complaint for total and partial indemnity. (Cf. Smith v. Lewis (1975) 13 Cal.3d 349, 360 [118 Cal.Rptr. 621, 530 P.2d 589, 78 A.L.R.3d 231] (failure to pursue arguable claims may constitute malpractice).) Aware that his settlement will not ordinarily prevent his participating in the litigation of the issues of damages and relative fault and that he might be held liable for further damages, a defendant contemplating settlement will rarely do so alone.
It has been suggested that statutes repudiating joint and several liability in comparative negligence cases are entitled to little, if any, weight in comparison to judicial opinions on the issue. However, in a democracy the laws enacted by the people’s elected representatives are entitled to great weight.
When the plaintiff is free of fault he is entitled to a joint and several judgment against each defendant in accordance with common law rule. The Li principle is inapplicable because there is simply no plaintiff fault for comparing with defendants’ fault.
In addition, when one defendant is held liable for the. acts of another on the basis of principles of vicarious liability, there should be no apportionment of liability because by definition one is liable for the acts of the other. (Ante, p. 587.) Apportionment between defendants should be denied even if the plaintiff is negligent, and in determining relative fault of plaintiff and defendants, the single negligent act for which both defendants are responsible should not be counted twice.
To illustrate, if plaintiff and the solvent defendant are equally at fault, the amount to be recovered will depend on the extent of fault of the insolvent defendant. If the *615insolvent defendant is 80 percent at fault, plaintiff will recover 90 percent of his loss but if the insolvent is only 10 percent at fault, recovery will be limited to 55 percent of the loss.
Existing rules should be continued as to nonnegligent plaintiffs.
In the instant case, plaintiff alleges defendants negligently conducted a motorcycle race. Defendant American Motorcycle Association alleges that plaintiff was negligent in causing the accident and that pláintiff’s parents negligently failed to supervise their minor child. Assuming that both plaintiff and defendant are successful in proving their allegations, the division of the loss between plaintiff, defendant, and the parents will require arbitrary allocation.