We must decide, under California law, whether an insurer for incidents of employee dishonesty properly refused to pay the coverage limit for the first two of three separate one-year policy periods, in connection with an employee’s fraudulent scheme that spanned all three periods.

I

Karen Kane, Inc. (“Kane”), a California corporation, is a Los Angeles-based apparel manufacturer. Rebanee Insurance Company (“Reliance”) is a Pennsylvania corporation authorized to sell insurance in the State of California. Reliance issued Kane a package of commercial insurance policies that included coverage for “employee dishonesty,” with a crime coverage limit of $250,000. Rebanee issued Kane three employee dishonesty policies for the following one-year policy periods: (1) December 1993 to December 1994 (“1993-94”); (2) December 1994 to December 1995 (“1994-95”); and (3) December 1995 to December 1996 (“1995-96”). Although Reliance issued Kane three separate policies, one for each period, the policies each employ the exact same policy language.

Norris Dantzler, one of approximately 225 Kane employees, began working in Kane’s trim department in 1987. In late 1992, Dantzler began to engage in a scheme to defraud Kane by having it pay for trim items that it did not order and did not receive. Dantzler was assisted in this scheme by confederates outside of Kane who controlled companies that were some of Kane’s current or former suppliers. Dantzler’s confederates would prepare falsified invoices for trim goods that Kane had supposedly ordered and received. In some cases, the amount of merchandise on the invoice had been falsified, so that Kane received some but not all of the listed merchandise; in other cases, the entire invoice was for merchandise never ordered nor received by Kane. Dantzler would then arrange to have an authorized Kane officer approve a check request for payment of the invoice, often “walking through” the request himself to avoid detection of the scheme. Finally, Dantzler’s co-conspirators would negotiate Kane’s checks at a check-cashing service. This scheme began in late 1992 and continued until May 1996, when it was discovered by Kane’s president, Lonnie Kane.

In June 1996, pursuant to the employee dishonesty insurance provided by Reliance, Kane filed separate Property Loss Notices with Reliance for the policy periods 1994-95 and 1995-96. Reliance assigned Russell LaDue to process Kane’s claims. In July 1996, Kane submitted to Reliance a Fidelity Proof of Loss in the amounts of $239,879.15 for 1995-96 and “at least $250,-000” for 1994-95. In August 1996, Kane submitted a Fidelity Proof of Loss in the amount of $484,978.07 for 1993-94. Kane subsequently reduced its claim for 1995-96 to $222,481.15 after collecting $17,398 from two of Dantzler’s co-conspirators. The parties have stipulated that Kane suffered losses in connection with Dantzler’s scheme of $434,978 in 1993-94, $767,298 in 1994-95, and $213,737 in 1995-96.

In a letter dated August 21, 1996, Kane stated to Reliance its view that A.B.S. Clothing Collection, Inc. v. Home Insurance Co., 34 Cal.App.4th 1470 , 41 Cal.Rptr.2d 166 (1995), required Reliance to pay the $250,000 coverage limit for each of three separate policy periods. LaDue responded that Reliance was rejecting any claim under the 1993-94 policy because “it is part of the loss of ... 1995-96,” citing for support the policy’s definition of the word “occurrence.” The policy defines “occurrence” as “all loss caused by, or involving, one or more employees, whether the result of a single act or series of acts.” Reliance informed Kane that under Reliance’s interpretation of the policy, “even though [the loss from Dantzler’s scheme] took place over a span of 3 years, it’s considered 1 occurrence and the insurance company would be liable for the limits of the policy during the policy period of 12-19-95 to 12-19-96.” Reliance also cited for support Paragraph B.9 of the Crime General Provisions (the “prior loss provision”), which provides as follows:

Kane subsequently sent a letter objecting to Reliance’s interpretation, relying once again on the A.B.S. decision. Reliance responded by letter arguing that the case was distinguishable because the policy in A.B.S. involved “different policy numbers,” while the policy at issue here “was renewed each year using the same policy number, and therefore, we feel this indicates a continuing policy and the insured would be limited to coverage for one policy period only, as this would be one occurrence under the terms of the policy.” Before the district court, however, Reliance conceded that each of the one-year policies it issued to Kane constituted a separate policy, rather than one part of a single continuous policy.

On October 31,1996, Reliance paid Kane $250,000 under the 1995-96 crime coverage, stating that $250,000 represented the “policy limits of this loss.” Kane accepted this payment but reserved the right to sue Reliance for breach of contract and wrongful denial of coverage.

On March 3, 1997, Kane filed suit against Reliance in California Superior Court for declaratory relief, breach of contract, and tortious breach of the covenant of good faith and fair dealing. Reliance removed the case to federal court on May 2, 1997. On cross-motions for summary judgment, the district court granted summary judgment in favor of Reliance, holding that the policy’s definition of “occurrence” and the prior loss provision clearly limited Kane’s recovery to the liability limit for the last of the three policies, for 1995-96. The district court dismissed all of Kane’s claims in a judgment entered February 18, 1998. Kane filed a timely notice of appeal.

II

The first issue presented is whether the district court erred in holding, under California law, that the policy’s definition of “occurrence” and its “prior loss” provision precluded any recovery by Kane under the first two of three insurance policies (for 1993-94 and 1994-95). Kane contends that the district court erred (1) by misapplying California law in interpreting the policy and (2) by misinterpreting the policy’s definition of the term “occurrence.” We address these contentions in turn.

A

Kane’s primary contention on appeal is that the district court erred through misapplying California law. More specifically, Kane argues that the district court should have followed the analysis of A.B.S. Clothing Collection, Inc. v. Home Insurance Co., 34 Cal.App.4th 1470 , 41 Cal.Rptr.2d 166 (1995).

As the district court recognized, federal courts sitting in diversity must apply the substantive law of California, as interpreted by the California Supreme Court. See Karen Kane, Inc. v. Reliance Ins. Co., No. CV 97-3295-WMB, 1998 WL 476454 , at *3 (C.D.Cal. Feb.17, 1998). Although A.B.S. is but a single decision by an intermediate appellate court, we note that no other appellate court in California has issued an opinion rejecting or expressing disagreement with its analysis. Furthermore, no evidence indicates, and neither party contends, that the California Supreme Court would decide the issues raised in AB.S. differently. We find A.B.S. to be a persuasive statement of California law.

The relevance of A.B.S. to this case is heightened by the numerous similarities in the fact patterns and policy language of the two cases. In each case, an employee stole funds from an employer over a period of several years during which the employer carried insurance against employee dishonesty issued by a single insurer. Furthermore, the policy issued to Kane by Reliance uses language that is identical in all relevant respects to the A.B.S. policy language. The district court in this case rested its decision on the policy’s definition of “occurrence” and its prior loss provision. The A.B.S. policy and the Reliance-issued policy define “occurrence” in the same way and contain prior loss provisions with identical wording. Thus A.B.S. and this case are, as a functional matter, indistinguishable.

The California Court of Appeal for the Second District stated the question presented in A.B.S. as follows: “When an employee embezzles funds from an employer over a period of years during which the employer carries insurance against employee dishonesty from the same insurer, may the employer recover up to the insurer’s limit of liability for each year in which the embezzlement occurs?” 34 Cal.App.4th at 1473 , 41 Cal.Rptr.2d 166 . The A.B.S. court explained that this question, one of first impression in California, could be answered by determining whether coverage is based on a series of separate contracts or on a single continuous contract:

Id. at 1473-74 , 41 Cal.Rptr.2d 166 .

The A.B.S. court then proceeded to determine whether the policies issued to A.B.S. by Home Insurance Company (“Home”) constituted multiple discrete contracts or a single continuous one. The court began its discussion with a heading stating the following principle: “An Insurer Seeking to Limit the Amount of Its Liability to the Insured for Losses Incurred During Successive Years of Coverage Must Show by Clear and Unambiguous Policy Language That the Parties Intended to Enter Into One Continuous Contract.” Id. at 1475 , 41 Cal.Rptr.2d 166 . Under this rule, an insurer mil be liable up to the policy limit for each separate period, unless it can show clear intent by the parties to enter into a single continuous insurance contract. As the A.B.S. court explained:

Id. at 1476 , 41 Cal.Rptr.2d 166 (citations omitted).

The A.B.S. court then conducted a fairly detailed examination of the policy’s specific provisions in order to assess whether the parties intended to enter into several separate contracts or a single continuous one. Two aspects of the A.B.S. court’s interpretation of the policy are relevant here. First, the court discussed whether the policy’s prior loss provision could establish an intent to enter into a continuous contract. The prior loss provision in A.B.S. provided as follows: “If any loss is covered (a) Partly by this insurance; and (b) partly by any prior cancelled or terminated insurance that we or any affiliate had issued to you ... the most we will pay is the larger of the amount recoverable under this insurance or the prior insurance.” Id. at 1479 , 41 Cal.Rptr.2d 166 . The A.B.S. court stated that this prior loss provision, even when read in combination with a “non-cumulation” provision, failed to show a clear and unambiguous intent to enter into a continuous contract. Id. at 1479 , 41 Cal.Rptr.2d 166 . Second, the A.B.S. court examined the policy’s definition of “occurrence” as “all loss ... whether the result of a single act or series of acts.” Id. at 1484-85 , 41 Cal.Rptr.2d 166 . The court found the above definition of “occurrence” to be ambiguous: “while defining ‘occurrence’ as ‘all loss’ suggests that there can be only one occurrence during the life of the insurance, the provision restricting liability ‘for any one occurrence’ suggests there could be more than one occurrence.” Id.

After reviewing these and other aspects of the policy, the AB.S. court concluded that “the policy provisions submitted on the motion for summary judgment do not clearly and unambiguously show Home and A.B.S. intended to enter into one continuous contract of insurance.... The policy provisions are at best ambiguous on these issues. These ambiguities must be construed [under principles of California insurance law] in favor of A.B.S.” Id. at 1485 , 41 Cal.Rptr.2d 166 . The A.B.S. court therefore reversed the trial court’s grant of summary judgment in favor of the insurer and remanded “for further proceedings consistent with the views expressed in our opinion.” Id.

The district court here was called upon to interpret a policy identical to the A.B.S. policy and to apply it to facts indistinguishable from those of AB.S. The policy issued by Reliance included the following prior loss provision, identical to that of the AB.S. policy: “If any loss is covered (a) [pjartly by this insurance; and (b) [p]artly by any prior cancelled or terminated insurance ... [t]he most we will pay is the larger of the amount recoverable under this insurance or the prior insurance.” Like the AB.S. policy, the Reliance policy defined “occurrence” as “all loss caused by, or involving, one or more ‘employees,’ whether the result of a single act or series of acts.” The definition of “occurrence” matters because under the policy, “[t]he most [Reliance] will pay for loss in any one occurrence is the applicable Limit of Insurance shown in the Declarations,” specifically, $250,000.

In contrast to the A.B.S. court, which held that both of these provisions were ambiguous and therefore had to be interpreted in favor of coverage, the district court here found both provisions unambiguous. The district court held that under the policy’s definition of “occurrence,” Dantzler’s multi-year conspiracy “falls squarely within the definition of occurrence provided by the policy: his fraud constituted a ‘series of acts’ that caused the insured a loss.” Kane, 1998 WL 476454 , at *5. Reading the policy’s definition of “occurrence” together or loss” provision, the district court held that “[b]ecause Dantzler’s scheme constitutes one ‘occurrence’ for purposes of the policy, the plaintiff is only entitled to recover under the current policy (1995-96), as the prior loss provision bars its claim under the two previous policies (1994-95 and 1993-94).” Id. at *7.

The problem with the above analysis is that the ambiguity found in the definition of “occurrence” by the A.B.S. court is present in this case as well. “Occurrence” could either (1) refer to the entire Dant-zler conspiracy (as a “series of acts,” namely, thefts), the position urged by Reliance; or (2) refer to each theft within the Dantzler conspiracy (as a “series of acts,” namely, the multiple steps involved in each theft). Similarly, with respect to the prior loss provision, the district court’s determination that the provision is unambiguous is inconsistent with the conclusion reached by the A.B.S. court when faced with the very same language. See 34 Cal.App.4th at 1479 , 41 Cal.Rptr.2d 166 . In light of the very strong similarities between AB.S. and this case, it would appear that the district court should have reached the same conclusion that the AB.S. court did: under California law, the provisions at issue are ambiguous and to be construed in favor of coverage.

The district court sought to reconcile the apparent inconsistency between its holding and that of the AB.S. court by observing that A.B.S. considered only whether policy provisions were ambiguous in the context of determining whether several one-year policies issued to the insured were separate or continuous. See Kane, 1998 WL 476454 , at *4-7. Here, in contrast, Reliance concedes that it issued three distinct policies to Kane. Thus, according to the district court and to Reliance, the A.B.S. court’s finding of ambiguity does not apply to this case.

While this argument has a certain appeal, a careful examination of AB.S. suggests that such a narrow reading of the opinion is unpersuasive. Although strictly speaking the A.B.S. court decided only whether the policies at issue in that case constituted separate policies or a single continuous policy, reading the A.B.S. opinion as a whole makes clear that the very reason the A.B.S. court concerned itself with this issue was to answer a more fundamental question, stated in the first paragraph of the opinion: “whether an insured is entitled to recover the limit of liability for each year a fidelity policy is in effect, when an employee’s dishonesty takes place in each year.” 34 Cal.App.4th at 1475-76 , 41 Cal.Rptr.2d 166 . The abstract, essentially academic question of “whether a series of one-year policies issued to the insured were separate or continuous” possessed no interest to the A.B.S. court independent of its bearing on the underlying question of the insurer’s liability.

The issue of separate or continuous contracts and the issue of whether or not an insurer’s liability is the limit for a single policy period are inextricably linked by the following general rule, stated in A.B.S.: if separate contracts are present, “the insurer is generally held liable up to its limit of liability for each policy period,” but if a single contract is involved, “the insurer’s liability is the amount stated in the contract regardless of the numbers of years involved.” Id. at 1476 , 41 Cal.Rptr.2d 166 . In deciding not to follow this rule as set forth in A.B.S., the district court focused on the fact that the A.B.S. court did not specifically hold Home liable up to the liability limit for each policy period. The district court stated that although the A.B.S. court noted that separate contracts “generally” trigger maximum liability under each contract, the California court did not “elaborate, however, on what circumstances might constitute an exception to this ‘general’ rule or trend.” 1998 WL 476454 , at *3. The district court regarded the California court of appeal’s connection between separate contracts and full liability, as opposed to a continuous contract and limited liability, as dicta.

Examination of the A.B.S. opinion suggests that this link might be better characterized as- an implied holding rather than mere dicta. Although the A.B.S. court focused on the issue of separate versus continuous contracts, the court was using this distinction as a rubric for analyzing the question presented in both A.B.S. and this case: “whether an insured is entitled to recover the limit of liability for each year a fidelity policy is in effect, when an employee’s dishonesty takes place in each year.” Id. at 1476, 41 Cal.Rptr.2d 166 . The A.B.S. court’s holding that the policies issued in that case were separate policies rather than one continuous policy cannot be considered in a vacuum. Rather, it must be recognized for what it was: a necessary step in that court’s determination of the insurer’s liability.

Although the district court here correctly noted that the A.B.S. court did not “specifically hold” the insurer liable, the procedural posture of A.B.S. prevented the court there from making such a specific holding. In AB.S., the insurer Home moved for and was granted summary judgment in the trial court; on appeal, the A.B.S. court had before it only the grant of summary judgment in favor of Home. Because A.B.S. had not moved for summary judgment, the court simply reversed the grant of summary judgment in favor of Home and “remanded to the trial court for further proceedings consistent with the views expressed in our opinion.” 34 Cal.App.4th at 1485 , 41 Cal.Rptr.2d 166 (emphasis added). In light of this remand, which precluded the insurer in A.B.S. from making on remand the arguments made by Reliance in this case, the A.B.S. discussion regarding liability cannot be dismissed as mere dicta.

B

Even if the language in A.B.S. connecting a continuous policy with full liability were dicta, Kane offers a second argument in support of its claim for coverage under the 1993-94 and 1994-95 policies. Kane contends that the term “occurrence” as defined in the policy is ambiguous with respect to whether it is temporally limited by each policy period, in which case Dant-zler’s ongoing fraudulent scheme would be recoverable as a separate “occurrence” within each period. If Kane’s argument is correct, then “occurrence,” as an ambiguous term, must under California law be construed in favor of coverage.

The insurance policy issued to Kane by Reliance provides that Reliance “will pay for loss of, and loss from damage to, Covered Property resulting directly from [employee dishonesty].” The policy places temporal limitations upon loss coverage: “we will pay only for loss that you sustain through acts committed or events occurring during the Policy Period.” As noted above, the policy defines “occurrence” as “all loss caused by, or involving, one or more ‘employees,’ whether the result of a single act or series of acts.” The policy further provides that “[t]he most [Reliance] will pay for loss in any one ‘occurrence’ is the applicable Limit of Insurance shown in the Declarations [$250,000].” As explained by the A.B.S. court, “[t]hese provisions create an ambiguity as to the extent of Home’s liability, because while defining ‘occurrence’ as ‘all loss’ suggests that there can be only one occurrence during the life of the insurance, the provision restricting liability ‘for any one occurrence’ suggests there could be more than one occurrence.” 34 Cal.App.4th at 1484-85 , 41 Cal.Rptr.2d 166 . Thus the policy is silent as to whether the term “occurrence” refers to “a single act or series of acts” within a single policy period or across multiple periods. If “occurrence” is construed as limited by policy period, then Dantzler’s approximately 150 individual acts of theft, spanning over three years, constitute three separate “series of acts,” one for each of the three policy periods and recoverable within each period as such.

In light of this ambiguity, Kane argues that the policy should be construed in favor of liability by reading “occurrence” as limited by each policy period. California case law suggests the plausibility of such an interpretation. See Stonewall Ins. Co. v. City of Palos Verdes Estates, 46 Cal.App.4th 1810, 1849 , 54 Cal.Rptr.2d 176 (1996). In Stonewall, the insured suffered loss arising out of “a continuous and repeated course of conduct of the City from 1971 to 1981 (namely, improper design and maintenance of a City storm drain ... ).” Id. at 1822-23 , 54 Cal.Rptr.2d 176 . A dispute arose among the City’s various insurers regarding proper apportionment of liability to the insured for a particular three year period. See id. at 1823 , 54 Cal.Rptr.2d 176 . Resolving this dispute required the court to interpret the term “occurrence” in the relevant policy. One policy at issue in Stonewall, issued by the Jefferson Insurance Company (“Jefferson”), defined “occurrence” as “an event or a continuous or repeated exposure to conditions which causes ... damage to property during the policy period that is neither knowingly nor intentionally caused.... ” Id. at 1834 , 54 Cal.Rptr.2d 176 . The Jefferson policy provided, however, that “[a]ll ... property damage arising out of a continuous or related exposure to substantially the same general conditions shall be considered one occurrence.” Id. (ellipsis in the original). Jefferson relied upon this provision in attempting to limit its liability to the $300,000 coverage limit applicable to a single policy period (rather than three separate policy periods). The Stonewall court rejected this argu ment, holding that any ambiguity in the term “occurrence” had to be construed in favor of liability. The court therefore read “occurrence” as limited by policy period, thus holding Jefferson liable for up to $900,000 in coverage (aggregating the $300,000 limit in each of the three policy periods). See id.

Following A.B.S. and Stonewall, we conclude that under California law, “occurrence” as defined under the Reliance policy is an ambiguous term with respect to temporal limitation and therefore must be construed in favor of liability. The district court thus erred in concluding that Kane could not recover under the 1993-94 and 1994-95 policies based on the policy’s definition of “occurrence” and the prior loss provision. The general rule of California law stated in A.B.S. is that an insurer which issues three separate policies for employee dishonesty is “liable up to its limit of liability for each policy period.” 34 Cal.App.4th at 1476 , 41 Cal.Rptr.2d 166 . Even if the A.B.S. discussion of liability were dicta, however, the term “occurrence” is ambiguous with respect to whether it is temporally limited by policy period. It therefore should have been interpreted in favor of coverage.

Ill

Reliance and Kane dispute whether the insurance policy’s one-year discovery rule would bar any recovery by Kane under the earliest of the three policies, for the 1993-94 policy period. The one-year discovery clause provides that Reliance “will pay only for covered loss discovered no later than one year from the end of the policy period.” The 1993-94 policy period ended on December 19, 1994; one year from the end of the policy period would have been December 19, 1995; and Kane did not discover Dantzler’s fraud until May 1996. Rebanee contends that Kane was not entitled to coverage under the 1993-94 policy because it discovered Dantzler’s fraud more than a year after the end of the 1993-94 policy period. Kane argues that (1) the one-year discovery rule is inconsistent with other policy provisions and (2)the one-year loss discovery period should have been tolled.

The language of the one-year discovery rule is quite plain and would appear to apply in straightforward fashion. Kane’s argument that the provision is “inconsistent” with other policy provisions, essentially because it limits the coverage provided for by those provisions, lacks merit. The district court correctly recognized that insurance companies are free to place limits upon the coverage of the policies they issue. See Indemnity Ins. Co. of North America v. California Stevedore & Ballast Co., 307 F.2d 513 , 518 n. 5 (9th Cir.1962) (“ ‘An insurance company has the right to limit the coverage of a pobey issued by it and when it has done so, the plain language of the limitation must be respected.’ ” (quoting Continental Casualty Co. v. Phoenix Constr. Co., 46 Cal.2d 423, 432 , 296 P.2d 801 (1956))). Simply because a general rule of coverage has certain clear exceptions does not mean that the exceptions should be read out of the policy as “inconsistent” with the rule.

Kane’s second argument against operation of the discovery-of-loss provision is that the discovery period is subject to tolling. Kane maintains that the one-year period provided for in the policy should be tobed until May 1996, when Kane discovered Dantzler’s fraud. This contention fails for two reasons. First, allowing for tolling would effectively read the one-year discovery provision out of the policy. The parties explicitly agreed that Reliance would “pay only for covered loss discovered no later than one year from the end of the policy penod” (emphasis added). If the one-year discovery period itself were tolled until discovery of loss, the discovery provision’s express limitation on coverage would be rendered meaningless.

Second, Kane’s argument that a tolling rule should override the language of the policy’s one-year discovery rule is not supported by California law. Kane argues for application of the tolling rule announced in Prudential-LMI Commercial Insurance v. Superior Court, 51 Cal.3d 674 , 274 Cal.Rptr. 387 , 798 P.2d 1230 (1990), to the one-year discovery period. Although Kane selectively quotes from that opinion to make it appear that Prudential sets forth a “delayed discovery rule” directly applicable to this case, examination of the opinion shows that its rule does not apply here.

Under California law, all fire insurance policies must provide “that no suit for recovery of any fire insurance claim shall be sustainable unless commenced within 12 months after the ‘inception of the loss.’ ” Prudential, 51 Cal.3d at 682 , 274 Cal.Rptr. 387 , 798 P.2d 1230 (quoting Cal. Ins.Code § 2071). In Prudential , the California Supreme Court considered “how to define the inception of a loss for purposes of triggering section 2071 when the loss occurs some time before any damage is discovered by the insured.” Id. at 684 , 274 Cal.Rptr. 387 , 798 P.2d 1230 (emphasis added). The court held that “ ‘inception of the loss’ should be determined by reference to reasonable discovery of the loss and not necessarily turn on the occurrence of the physical event causing the loss.” Id. at 686 , 274 Cal.Rptr. 387 , 798 P.2d 1230 .

In Prudential> then, the California Supreme Court simply interpreted the phrase “inception of the loss” in such a way so as to toll the period in between actual occurrence of a loss and when the loss became reasonably discoverable by the insured. The “inception” time of a loss is irrelevant, however, under the one-year discovery clause at issue here, which focuses on when a loss is “discovered.” Thus Prudential has no direct bearing upon and cannot override the plain language of Reliance’s discovery-of-loss provision. In light of the very significant differences between the policy language in Prudential and the policy language at issue here, Kane’s reliance on Prudential is misplaced.

California law supports application of the express language of the policy’s one-year discovery rule. In Admiralty Fund v. Peerless Insurance Co., 143 Cal.App.3d 379, 384 , 191 Cal.Rptr. 753 (1983), a California court of appeal made the following observations: “California has long recognized the validity of discovery of loss provisions in fidelity insurance policies. Generally, the courts have strictly enforced such provisions so that neither difficulty in discovering insured losses nor employee concealment excuse[s] the insured’s performance.”

Admiralty Fund presents facts much closer to those of the current case than Prudential. Admiralty Fund involved interpretation of a true discovery-of-loss provision quite similar to the provision at issue here, under which a loss was covered “only if discovered not later than one year from the end of the Policy Period.” Id. at 384 n. 3, 191 Cal.Rptr. 753 (emphasis added). After noting that such provisions are generally enforceable, see id., the Admiralty Fund court stated that the cases upholding such provisions “differ from the present situation of adverse domination and control, in which the very persons that control a company also perform the wrongdoings.” Id. at 384-85 , 191 Cal.Rptr. 753 . In such a case, explained the court, “there is no one to oversee those in charge, and so the concepts of diligence in uncovering an insurable loss and employee concealment are inapposite.” Id. Thus, in light of the possibility that “the dishonest president [of Admiralty Fund] and other high ranking officers controlled the Fund’s operations to such an extent as to preclude discovery [of their wrongdoing],” the trial court there should have considered the appropriateness of tolling the discovery period. Id. at 389 , 191 Cal.Rptr. 753 .

The Admiralty Fund opinion thus makes clear that discovery of loss provisions are generally valid and strictly enforceable, except in situations of “adverse and domination and control.” Id. at 384-85 , 191 Cal.Rptr. 753 . In the present ease, the facts stipulated to and alleged by Kane cannot, as a matter of law, support a finding that Dantzler exercised “adverse domination and control” over Kane. Unlike the president of the Admiralty Fund, Dantzler — not a corporate officer of Kane, but merely one of 225 employees — did not control the company. Nor did Dantzler function free of oversight by others. Rather, as the district court found, “the very nature of Dantzler’s scheme required review of his work by other employees — i.e., he needed other employees’ approval before checks could be issued.” In light of these facts, the district court here correctly held that under California law, the one-year discovery provision unambiguously barred Kane from recovering under the earliest of the three policies, for the 1993-94 pohcy period.

IV

As a final matter, we address Kane’s claim for breach of the covenant of good faith and fair dealing. Although Reliance’s reading of the insurance pohcy was ultimately incorrect in part, we find that Rehance’s interpretation was a reasonable one, noting that the district court found in Rehance’s favor. Because Reliance did not act unreasonably or tortiously in denying coverage, we affirm the dismissal of Kane’s bad faith claim. See Lunsford v. American Guarantee & Liability Ins. Co., 18 F.3d 653, 656 (9th Cir.1994) (applying California law and reversing the district court’s summary judgment that there was no insurance coverage under the pohcy, but holding as a matter of law that the insurer did not act in bad faith in denying coverage).

V

In sum, we hold that under California law Kane is entitled to payment up to the $250,000 pohcy limit under the second of the three insurance policies, for the 1994-95 period. The policy’s definition of “occurrence” is ambiguous and does not bar liability under the first two policies, for 1993-94 and 1994-95. The one-year discovery clause, however, independently precludes recovery under the pohcy for the 1993-94 period. Although Reliance incorrectly denied coverage under the 1994-95 pohcy, Reliance acted reasonably and did not violate the covenant of good faith and fair dealing in doing so. We therefore affirm in part and reverse in part the district court’s grant of summary judgment and remand for proceedings consistent with this opinion.

AFFIRMED in part, REVERSED in part, and REMANDED.

. We review a district court's grant of summary judgment de novo. See Newbery Corp. v. Fireman’s Fund Ins. Co., 95 F.3d 1392 , 1398 (9th Cir.1996). . In addition to finding support in case law, reading "occurrence" as limited by each policy period makes economic sense. If "occurrence” is interpreted as not limited by each policy period, Kane would have to pay for higher and higher coverage limits simply in order to maintain the same nominal coverage level for the type of loss it actually suffered. Under Reliance's reading of the policy, Kane essentially forfeited the $250,000 in coverage it purchased for the 1994-95 period simply by virtue of renewing for the 1995-96 period with the same $250,000 limit. Such an irrational arrangement would be inconsistent with the objectively reasonable expectations of the parties, which are the touchstone for interpreting insurance contracts under California law. . Kane also argues that if the one-year discovery rule does have the meaning that Reliance gives to it, then it should not be enforced because Reliance failed to give adequate warning to Kane of the provision’s limitation on coverage. This argument cannot be sustained in light of the following warning, prominently located at the beginning of the policy: "Various provisions in this policy restrict coverage. Read the entire policy to determine rights, duties and what is or is not covered.” Simply because Kane may not have subjectively anticipated that its recovery would be limited by the one-year discovery rule is irrelevant to the insured's objectively reasonable expectation of coverage, based on the clear meaning of the policy provisions. . The Prudential court also held that the one-year period in which the insureds were able to file suit under the policy was tolled during the period after the insureds submitted a claim for coverage to their insurer and before their claim was ultimately denied. This equitable tolling rule is irrelevant to the present action because unlike the plaintiffs in Prudential , Kane did not submit a timely claim, filing its claim for coverage more than a year after the 1993-94 policy period ended.