In this case, the majority addresses three issues concerning the rights and duties of insurer and insured under a standard commercial general liability (CGL) insurance policy when the insured has been sued for progressive property damage or personal injury caused by toxic pollution. In my view, the majority decides only one of these issues correctly.
The first issue that the majority addresses is whether site investigation costs—that is, costs incurred to determine the origin, nature, and extent of toxic pollution and the most appropriate remedial measures—are defense costs that the insurer must incur in fulfilling its duty to defend. Starting from the premise that defense costs and indemnity costs are mutually exclusive, the majority concludes that site investigation costs are defense costs rather than indemnity costs when the site investigation is conducted “within the temporal limits of the insurer’s duty to defend” and “amount[s] to a reasonable and necessary effort to avoid or at least minimize liability,” provided that the claimed expenses are “reasonable and necessary for that purpose.” (Maj. opn., ante, at pp. 60-61.)
*78Instead of starting from the premise that defense costs and indemnity costs are mutually exclusive, I would begin by acknowledging that site investigation may serve both to avoid or minimize liability and to discharge a cleanup obligation, and therefore a particular site investigation cost may be both a defense cost and an indemnity cost. Next, I would separately consider the situations of insureds who did not pollute, insureds who polluted but did not expect or intend the resulting harm, and insureds who polluted and did expect or intend the resulting harm.
In the first situation, in which it is ultimately determined, by judgment or settlement, through site investigation or otherwise, that the insured has no liability for toxic pollution, the absence of liability means-that there are no indemnity costs. In this situation, all site investigation costs reasonably incurred as part of the ultimately successful effort to avoid liability should be treated as defense costs that the insurer must pay under its duty to defend. In the second situation, in which it is ultimately determined that the insured is a toxic polluter but one who did not expect or intend the resulting harm, the duty to defend requires the insurer to pay site investigation costs reasonably incurred to avoid or minimize liability, but if those same costs also serve to discharge the insured’s cleanup liability, then they also constitute indemnity costs that the insurer may credit against the liability limits of the policy. In the third situation, in which it is ultimately determined that the insured is a toxic polluter who expected or intended the resulting harm, the public policy against indemnity for willful wrongs must be considered in deciding whether the insurer is obligated to pay site investigation costs. To the extent that site investigation serves to discharge the insured’s cleanup liability, this public policy requires that the insured rather than the insurer bear those cleanup costs.
The second issue that the majority addresses is whether, when the insurer has incurred costs in the defense of what the majority terms a “mixed claim,” meaning a claim having parts that are potentially covered and parts that are not potentially covered, the insurer may obtain reimbursement from the insured for defense costs that may be allocated solely to parts of the claim that are not potentially covered. The majority decides that an insurer may obtain reimbursement from the insured for such costs, treating this issue as controlled by this court’s recent decision in Buss v. Superior Court (1997) 16 Cal.4th 35 [65 Cal.Rptr.2d 366, 939 P.2d 766]. I dissented in Buss (id. at p. 62) and would not further extend the scope of that misguided decision, which ignored clear policy language and standard principles of contract law.
The third issue concerns equitable allocation of defense costs among insurers having defense obligations that overlap because each insurer afforded coverage during part of the time when a progressive property damage *79or personal injury was occurring. Specifically, the issue is whether, in the process of equitably allocating defense costs among insurers having overlapping defense obligations, an insured who lacked coverage during part of the period in question may be treated as its own insurer and on this basis equitably allocated a portion of the defense costs. The majority concludes that in this situation those insurers having a duty to defend are required to bear all defense costs and that no portion of the defense costs may be allocated to the insured. I agree and concur in this portion of the majority opinion.
I
Since 1951, Aerojet-General Corporation (Aerojet) has operated a large facility near Sacramento to develop and produce missile and rocket motors under government contracts. Between 1956 to 1984, the time period relevant here, Aerojet purchased liability coverage under a series of CGL policies issued by various insurers. Between 1976 and 1984, Aerojet purchased “cash flow” or “fronting” policies from Insurance Company of North America (INA). These policies contained endorsements that shifted to Aerojet the ultimate cost of both indemnification and defense.
In 1979, it was discovered that the groundwater beneath and around Aerojet’s Sacramento facility had been contaminated with trichloroethylene (TCE), a chemical solvent that Aerojet had used in large quantities. Discovery of the TCE pollution led the federal and state governments and various private parties to sue Aerojet, which sought indemnification and defense from its liability insurers. The insurers (except INA) provided a defense but reserved their rights to dispute indemnity coverage.
Several insurers commenced this action by filing a complaint for declaratory relief to clarify their obligations under the CGL policies they had issued to Aerojet. Aerojet cross-complained against all the insurance companies that had issued CGL policies to it between 1956 and 1984. The insurers’ complaint eventually was dismissed without prejudice, and the case went to trial on Aerojet’s cross-complaint. The action was tried in three phases. In the first phase, the trial court construed the relevant disputed language of the insurance policies. In the second phase, the jury applied the policy language as construed by the trial court to the evidence presented. Specifically, it applied the standard CGL policy language limiting the insurers’ obligations to harm that was “neither expected nor intended from the standpoint of the insured,” and the jury concluded that there was no indemnity coverage for the damages occasioned by the TCE pollution because Aerojet had expected or intended the harm caused by that pollution.
*80Before the third phase began, the parties entered into a detailed stipulation. They agreed that the insurers had paid or would pay all ordinary defense costs, such as attorney fees and expert witness fees, through the final disposition of any appeal Aerojet would take from the judgment to be entered in the action. They also agreed that in the third phase the parties would litigate only whether site investigation expenses were defense costs, the insurers would not seek reimbursement of any part of the ordinary defense costs, and Aerojet would not seek “bad faith” damages. Finally, they agreed that if the case were to be retried following a reversal on appeal of the judgment that would be entered after the conclusion of the third phase, then upon retrial the insurers would be free to seek reimbursement for defense costs from Aerojet, and Aerojet in turn would be free to seek “bad faith” damages from the insurers.
During the third phase of the trial, the jury determined that costs of government-mandated site investigations were not defense costs, and therefore the insurers were not required to pay these costs to discharge their duty to defend. The trial court entered judgment against Aerojet on its cross-complaint. Aerojet appealed, raising issues relating to each of the three phases of the trial.
The Court of Appeal concluded that although the trial court had committed certain errors in the first and second phases of the trial, Aerojet had not been prejudiced by these errors in light of the overwhelming evidence that Aerojet knew both that TCE was a highly toxic and harmful pollutant and that its disposal practices were causing TCE pollution of the groundwater. Thus, the Court of Appeal upheld the jury verdict at the second phase that, because Aerojet had expected or intended the harm caused by the TCE pollution, the insurers had no indemnity obligation.
As for the third phase, the Court of Appeal reached these conclusions: (1) The jury’s finding that none of the site investigation expenses were defense costs was prejudicially affected by erroneous instructions; and (2) liability for defense costs should be equitably apportioned among the various insurers, and in this apportionment Aerojet should also bear á reasonable share because it chose to be self-insured for defense costs, by means of the INA “fronting” policies, for part of the time during which the harm was occurring. The Court of Appeal reversed the judgment as to the third phase only and remanded for a limited new trial.
Aerojet petitioned this court for review, raising several issues. Three of the insurance carriers also petitioned for review, raising only the issue *81whether a CGL insurer’s defense obligation includes payment of site investigation costs. This court granted both petitions but ultimately limited the issues on review to those that the majority addresses. This court has declined to address any issue concerning the determination that, because Aerojet expected or intended the harm caused by the TCE pollution, the insurers have no indemnity obligation. Accordingly, that determination is no longer subject to appellate review and will become final.
II
The standard CGL policy defines the insurer’s indemnity obligation as the duty to pay “all sums that the insured becomes legally obligated to pay as damages” as a result of personal injury or property damage or other covered risk, provided that the damages are to compensate for harm that was “neither expected nor intended from the standpoint of the insured.” In AIU Ins. Co. v. Superior Court (1990) 51 Cal.3d 807, 814 [274 Cal.Rptr. 820, 799 P.2d 1253], we held that in this standard language the term “damages” includes cleanup and other “response” costs incurred under the Comprehensive Environmental Response, Compensation, and Liability Act of 1980 (CERCLA) (42 U.S.C. § 9601 et seq.) and related state and federal environmental laws. Thus, the term “damages” is not limited to money that the insured must pay to an injured third party under a liability judgment, but includes also money that the insured must pay to comply with a legal obligation for cleanup of toxic pollution.
Apart from its duty to indemnify, an insurer under a CGL policy has a duty to defend the insured in any “suit” seeking damages potentially within the policy’s indemnity coverage. (See Buss v. Superior Court, supra, 16 Cal.4th 35, 62-63 (dis. opn. of Kennard, J.).) It is settled, and not disputed here, that an insurer’s duty to defend includes a duty to investigate the factual allegations of a third party’s claim against the insured whenever the investigation constitutes a reasonable effort to avoid or reduce the insured’s liability.
In a standard CGL policy, indemnity coverage is subject to dollar limits specified in the policy, but defense coverage is not subject to any specified dollar limit. Otherwise stated, the policy’s monetary limits have no application to defense costs.
Under CERCLA, once a federal agency identifies a suspected polluter (referred to in CERCLA as a “potentially responsible party” or PRP), the agency may require the PRP to conduct a site investigation (which CERCLA *82calls a “remedial investigation/feasibility study” or RI/FS) to determine the extent of the pollution and the most feasible method of cleaning it up. If the PRP does not comply with the agency’s request to perform an RI/FS, the agency will perform the RI/FS itself and charge the cost to the PRP. Generally, however, it is advantageous for the PRP to perform the RI/FS itself, because the results of the study are critical in determining whether, and if so how much, the PRP will eventually have to pay for the cleanup.
Because the results of the RI/FS can be used to demonstrate that there is no toxic pollution, that a PRP is not the source of the pollution, or that the pollution is less severe or less extensive than claimed, or to demonstrate the feasibility of cleanup methods less costly than those the agency might otherwise have demanded, Aerojet and other insureds claim that the costs of the RI/FS are entirely defense costs that their insurers must pay. If the costs are treated only as defense costs, they do not count against the policy limits. This is critical in a case like this one, in which some $26 million has been spent on site investigation, including the RI/FS.
The majority states that an insurer, to discharge its duty to defend, “must undertake reasonable and necessary efforts to avoid or at least minimize liability” and for this purpose “must incur reasonable and necessary costs.” (Maj. opn., ante, at p. 60.) From these simple and indisputable propositions, the majority concludes that site investigation expenses are defense costs, and not indemnity costs, whenever the site investigation is “conducted within the temporal limits of the insurer’s duty to defend” and amounts “to a reasonable and necessary effort to avoid or at least minimize liability,” provided only that the particular expense incurred is “reasonable and necessary for that purpose.” (Id. at pp. 60-61.)
The majority states that this court’s decision in AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d 807, is “[n]ot to the contrary” and stands only for the proposition that certain site investigation expenses may be indemnity costs, • not for the proposition that site investigation costs may never be defense costs. (Maj. opn., ante, at p. 61.)
In a footnote, the majority states that defense costs and indemnity costs are mutually exclusive and that the same cost cannot be both an indemnity cost and a defense cost. (Maj. opn., ante, at p. 61, fn. 13.) The majority distinguishes the two by stating that indemnity costs “arise after the insured’s liability is established and as a result thereof’ whereas defense costs “arise before the insured’s liability is established and apart therefrom.” (Ibid.)
*83Two basic defects permeate the majority’s analysis of this issue.
First, the majority is wrong in its assertion that defense costs and indemnity costs are mutually exclusive and may be readily distinguished by determining whether the cost was incurred before or after the insured’s liability was established. The majority assumes that in every situation that may arise under a CGL policy, there is a single point in time when the insured’s liability is fixed, so that expenses incurred after that point cannot be defense costs (because there is no longer anything to defend) while expenses incurred before that point cannot be indemnity costs (because there was not yet a liability to discharge). But defense and indemnity costs are not so easily separated. In particular, it frequently happens that an insured will admit the fact of liability but dispute the amount or extent of that liability. Here, for example, Aerojet conceded at an early stage that it was the source of the TCE pollution around its Sacramento facility, and it agreed in principle to pay the costs of cleanup; thereafter, the dispute shifted to issues such as the extent of the pollution and the most cost-effective method of cleanup. After conceding the fact of liability, and while disputing the extent of liability, an insured may incur costs that are defense costs as well as indemnity costs because they are incurred for diagnostic or other services that operate both to reduce and to discharge liability. In the context of a CERCLA enforcement action, an RI/FS may serve these dual purposes. In the context of a personal injury claim, the cost of a medical diagnostic procedure may likewise be both a defense cost and an indemnity cost because the procedure may both show that the injury is less serious (and thus the insured’s potential liability is less costly) than claimed and also constitute a necessary and integral part of the treatment for the injury. In brief, a single expense may qualify as an indemnity cost because it serves to partially discharge the insured’s liability and as a defense cost because it assists, or could assist, in limiting the scope of the insured’s liability.
The majority’s “temporal limits” approach, under which defense costs and indemnity costs are mutually exclusive and the categorization of a particular cost is determined by reference to whether it was incurred before or after liability was fixed, cannot be applied successfully to the complex situations that arise under CGL policies. Toxic pollution cases are particularly resistant to this approach because the actual monetary cost of the polluter’s liability is rarely fixed in advance, but is usually determined retrospectively, after the cleanup is finished. If the tab is totaled only at the end, there is room to argue that costs incurred at every step of the process are defense costs because they may reduce subsequent cleanup costs by demonstrating that there are cheaper cleanup methods than those advocated by the government, that the methods used have been effective, or that the pollution is actually *84less severe than thought. There is also room to argue that none of the costs are indemnity costs that the insurer may charge against the policy limits because all were incurred before the full extent of the insured’s liability was established.
The second global defect in the majority’s approach is that it completely ignores the public policy against indemnity coverage for intentional torts. This policy is embodied in Insurance Code section 533, which provides that an insurer “is not liable for a loss caused by the wilful act of the insured.” (See J. C. Penney Casualty Ins. Co. v. M.K. (1991) 52 Cal.3d 1009, 1019-1021 [278 Cal.Rptr. 64, 804 P.2d 689].) As the majority recognizes, the standard CGL policy language limiting coverage to harm “neither expected nor intended from the standpoint of the insured” implements Insurance Code section 533. As the majority puts it, “If specified harm is ‘expected’ or ‘intended’ by the insured, it is effectively caused by a ‘wilful act’ within the meaning of Insurance Code section 533, and hence outside of coverage.” (Maj. opn., ante, at p. 49.) Despite its acknowledgment of this important public policy, the majority gives the policy no consideration in deciding whether site investigation costs should be categorized as defense costs, indemnity costs, or both. The majority’s disregard for this policy is remarkable given the determination in this very case that Aerojet expected or intended the harm caused by the TCE pollution. By requiring Aerojet’s insurers to foot the bill for all site investigation that might have proved useful to the defense effort, without regard to whether the site investigation was also necessary to remedy the harm caused by the intentional toxic pollution, the majority effectively compels at least partial indemnity coverage for an intentional tort.
I propose a different approach. I would begin by asking whether the insured is a nonpolluter, a polluter who neither expected nor intended harm, or a polluter who expected or intended harm. I would treat each of these situations differently.
If it is ultimately determined that the insured did not pollute and thus has no cleanup obligation, there can be no overlap between defense costs and indemnity costs because the insured has no liability that could trigger an insurer’s indemnification duty. In this situation, site investigation costs reasonably incurred for a proper defense purpose are defense costs that the insurer must pay, without limitation as to amount, under its duty to defend. As the majority and I agree (maj. opn., ante, at pp. 76-77, fn. 29), this conclusion follows from the standard CGL policy language.
If it is ultimately determined that the insured is responsible for toxic pollution, and thus also that the insured has a cleanup obligation, there may *85be an overlap between defense costs and indemnity costs because site investigation may serve both to reduce and to discharge the insured’s cleanup liability. But the consequences of this overlap cannot be explained without, distinguishing between, on the one hand, an insured who expected or intended the harm resulting from the pollution, and, on the other hand, an insured who did not expect or intend that harm.
If the insured neither expected nor intended the harm caused by the pollution, the insurer under a standard CGL policy has both a duty to defend and a duty to indemnify. The defense obligation is not subject to policy limits; the indemnity obligation is. In this situation, if a particular site investigation activity serves both a proper defense purpose and to discharge the insured’s cleanup liability, the reasonable cost of that activity is both a defense cost and an indemnity cost. Because it is a defense cost, the insurer must pay the site investigation expense without regard to policy limits (in other words, no amount of such payments can exhaust the duty to defend). Because it is also and equally an indemnity cost, however, the insurer may credit the payment against the policy’s limit on indemnity coverage, thereby reducing its obligation to pay other indemnity costs. These consequences necessarily follow from the language of the standard CGL policy.
The majority disputes this conclusion, apparently based on its stubborn insistence that site investigation costs may be characterized as either defense costs or indemnity costs, but not both. (Maj. opn., ante, at pp. 76-77, fn. 29.) The majority cites no provision of the standard CGL policy precluding dual characterization of costs or precluding the insurer from counting such costs dual-purpose costs against the limits of indemnity coverage. For example, the standard CGL policy might provide, but does not, that indemnity expenses count against the policy limits except when they also serve a proper defense purpose. Absent limiting language of this sort, the unambiguous policy must be construed to provide that insurers may count against the policy limits all indemnity expenses regardless of whether they also serve a proper defense purpose.
The remaining situation is the one at issue here, in which the insured expected or intended the harm resulting from the insured’s own acts of toxic pollution. Because the insured has a cleanup obligation, there is a potential for overlap of defense and indemnity costs because site investigation may serve both to determine and to discharge the insured’s cleanup obligation. But in this situation there is no indemnity coverage, both by virtue of the standard CGL policy exclusion for harm that the insurer expected or intended and by operation of Insurance Code section 533’ s prohibition against *86indemnity for losses caused for willful acts. Thus, the insured may not require the insurer to pay costs necessary to discharge the cleanup liability. Nevertheless, the insurer is obligated to provide the insured with a defense until it is established that the insured expected or intended the harm. (See Horace Mann Ins. Co. v. Barbara B. (1993) 4 Cal.4th 1076 [17 Cal.Rptr.2d 210, 846 P.2d 792].)
If a particular site investigation activity is useful both to reduce and to discharge the insured’s cleanup liability, the reasonable cost of that activity is both a defense cost and indemnity cost. Because it is a defense cost, the insurer is obligated under the standard CGL insurance policy to pay it if it was incurred before the determination was made that the insured expected or intended the harm. But because it is also an indemnity cost, the insurer is prohibited by Insurance Code section 533 from paying it, regardless of when it was incurred. This conflict in the insurer’s obligations—under which it must both pay and not pay the same cost—must be resolved. I would resolve it in favor of enforcing Insurance Code section 533 and relieving the insurer from any contractual obligation it might otherwise have to pay these dual-purpose costs as defense costs. This analysis is based on the standard CGL policy language and on Insurance Code section 533. The resolution of the conflict—a conflict that the majority fails even to acknowledge—is designed to best implement Insurance Code section 533.
Of course, site investigation costs are often incurred before it has been established that the insured expected or intended the harm resulting from the toxic pollution. So long as there is a possibility (because it has not been established that the exclusion for willful harm applies) that the insurer has a duty to indemnify for liability resulting from the pollution, the insurer must pay, as incurred, the reasonable costs of site investigation that serves, or could serve, a proper defense purpose. (See Gray v. Zurich Insurance Co. (1966) 65 Cal.2d 263, 275-279 [54 Cal.Rptr. 104, 419 P.2d 168].) Upon a determination that the insured expected or intended the harm resulting from the pollution, the insured should be required to reimburse the insurer for the cost of all site investigation that served to discharge the insured’s cleanup obligation.
The majority asserts that insofar as this reimbursement serves to relieve the insurer of a contractual obligation to pay costs incurred in the defense of its insured against a “wilful act” claim, it is inconsistent with prior decisions of this court. (Maj. opn., ante, at pp. 76-77, fn. 29.) But none of the decisions the majority cites considered the problem of dual-purpose costs, and none required an insurer to pay indemnity costs for harm caused by a willful act of the insured merely because the same costs might also be characterized as *87defense costs. Indeed, the majority here takes the unprecedented step of establishing a nonstatutory exception to Insurance Code section 533’s prohibition against indemnity for harm caused by a willful act of the insured.
Although the views I have stated provide a somewhat complicated answer to what may have seemed a simple question, they not only follow logically from the standard CGL policy language and from Insurance Code section 533, but also accommodate the various public and private interests implicated. when an alleged polluter with CGL coverage incurs costs for site investigation.
Ill
On the second issue, the majority concludes that a CGL insurer may obtain reimbursement from the insured for defense costs that may be allocated solely to parts of claims that are not potentially covered under the policy. (Maj. opn., ante, at p. 69.) The majority states that this conclusion is logically compelled by this court’s decision in Buss v. Superior Court, supra, 16 Cal.4th 35. (Maj. opn., ante, at p. 69.)
Because the majority offers nothing new on this point, relying instead on its analysis in Buss v. Superior Court, supra, 16 Cal.4th 35, my dissent to that decision (id. at p. 62) explains why the majority’s reasoning is unpersuasive here. Briefly, the standard CGL policy uses language obligating the insurer to defend any “suit” that potentially seeks covered damages, and it contains no provision obligating the insured to reimburse the insurer for the cost of defending particular claims within such a suit that are not potentially covered. Applying standard principles of contract interpretation to the policy language, the conclusion is inescapable that when an action includes both claims that are potentially covered and claims that are not potentially covered, or claims that are potentially covered in some parts and not in others, the insurer is contractually obligated to defend the entire suit.
As it did in Buss v. Superior Court, supra, 16 Cal.4th 35, the majority ignores the policy language obligating the insurer to defend the whole of any “suit” asserting a potentially covered claim, without reserving any right of reimbursement for costs that may be attributable to claims, or parts of claims, that are not potentially covered. Indeed, the majority’s holding is even more troubling than its holding in Buss because it requires acceptance of the proposition that individual claims may be subdivided into parts, each of which can then be characterized as potentially covered or not potentially covered. This minute dissection is likely to be a litigation nightmare for trial *88courts. Although the decision in Buss is now the law, there is no reason to extend it further, and every reason to limit it to the greatest possible extent.
IV
On the third issue, the majority holds that although insurers having overlapping defense obligations may equitably apportion defense costs among themselves, an insured is not required to make a contribution together with the insurers, even when the insured lacked insurance during part of the time that the progressive loss was occurring. (Maj. opn., ante, at p. 72.) I concur in the reasoning of the majority on this issue only.
Conclusion
Here, the jury determined that Aerojet’s site investigation costs were not defense costs that Aerojet’s CGL insurers were required to pay to discharge their duty to defend Aerojet. Because Aerojet has been determined to have expected or intended the harm cause by the TCE pollution, because public policy precludes indemnification for harm that the insured expected or intended, and because site investigation was a necessary step in discharging Aerojet’s cleanup obligation, I would not require Aerojet’s insurers to pay the site investigation costs but would instead uphold the jury determination on this issue. I dissent from the majority opinion to the extent it reaches a contrary conclusion.
Because an insurer’s contractual defense obligation under the standard language of a CGL policy runs to the whole of any action seeking potentially covered damages, not to individual claims or parts of claims, I dissent from the majority’s holding that Aerojet must reimburse its insurers for any defense costs that may be allocated solely to parts of claims not potentially covered.
I concur in the majority’s holding that in the equitable apportionment of defense costs among insurers having overlapping defense obligations, no share of these costs may be allocated to Aerojet on the theory that it was its own insurer under the INA fronting policies.
I would reverse the judgment of the Court of Appeal and remand the matter to that court with directions to affirm the judgment of the superior court.