Opinion
LUCAS, C. J.
We granted review to consider the Court of Appeal’s reversal of a directed verdict of coverage in favor of Jack and Rita Garvey (hereafter plaintiffs). We sought to resolve some of the confusion that has arisen regarding insurance coverage under the “all risk” section of a homeowner’s insurance policy when loss to an insured’s property can be attributed to two causes, one of which is a nonexcluded peril, and the other an excluded peril.
In recent years, some courts have misinterpreted and misapplied our decisions in Sabella v. Wisler (1963) 59 Cal.2d 21 [27 Cal.Rptr. 689, 377 P.2d 889], and State Farm Mut. Auto. Ins. Co. v. Partridge (1973) 10 Cal.3d 94 [109 Cal.Rptr. 811, 514 P.2d 123]. In so doing, they have allowed coverage in first party property damage cases under our holding in Partridge by inappropriately using the Partridge concurrent causation approach [399]*399as an alternative to Sabella's efficient proximate cause analysis.1
This extension of the analysis in Partridge, a third party liability case, allows coverage under a first party property insurance policy whenever a covered peril is a concurrent proximate cause of the loss, without regard to the application of specific policy exclusion clauses.2 Such reasoning ignores the criteria set forth in Insurance Code sections 530 and 532,3 the relevant analysis in Sabella and the important distinction between property loss coverage under a first party property policy and tort liability coverage under a third party liability insurance policy. Indeed, because a covered peril usually can be asserted to exist somewhere in the chain of causation in cases involving multiple causes, applying the Partridge approach to coverage in first party cases effectively nullifies policy exclusions in “all risk” homeowner’s property loss policies, thereby essentially abrogating the limiting terms of insurance contracts in such cases. We cannot believe Partridge intended such a sweeping result in first party property loss cases. To the contrary, as we explain below, we must put Partridge in its proper perspective, i.e., that decision should be utilized only in liability cases in which true concurrent causes, each originating from an independent act of negligence, simultaneously join together to produce injury. Therefore, as will appear, we conclude this case should be remanded to the Court of Appeal with directions to remand to the trial court for a jury determination of causation pursuant to Sabella, supra, 59 Cal.2d 21.
I.
Facts
Plaintiffs bought their house in the mid-1970’s. In 1977, plaintiffs purchased from State Farm Fire and Casualty Company (hereafter defendant) an “all risk” homeowner’s policy of insurance which was in effect at all times relevant. Section I of the policy in question provided coverage for “all risks of physical loss to the property covered” except as otherwise excluded or limited. Losses excluded by this portion of the policy included those “caused by, resulting from, contributed to or aggravated by any earth movement, including but not limited to earthquake, volcanic eruption, landslide, mudflow, earth sinking, rising or shifting,” and losses caused [400]*400“by . . . settling, cracking, shrinkage, bulging or expansion of pavements, patios, foundations, walls, floors, roofs or ceilings. . . .”
In August 1978, plaintiffs noticed that a house addition, built in the early 1960’s, had begun to pull away from the main structure. They also discovered damage to a deck and garden wall. There ensued numerous phone calls, letters, meetings and investigations as plaintiffs tried to determine from defendant whether the damage was covered by their homeowner’s property insurance policy.
In October 1979, after receiving from its counsel an opinion that the loss was not covered, defendant notified plaintiffs by letter that the “policy excludes coverage for the loss herein. Normally, such a denial of coverage would leave you to your remedies, [j|] However, because the company wishes to resolve the coverage issue in an atmosphere free from extraneous matters such as bad-faith and class action issues, the company is prepared to advance you the claimed sum of $11,550 subject to a reservation of rights as authorized by Johansen v. CSAA, 15 Cal.3d 9. . . .” Under the agreement proposed, defendant would make the advance and file a declaratory relief action on the issue of coverage; plaintiffs would pay back the advance if the court ruled in defendant’s favor, would waive “any claim of consequential or punitive damages arising out of any allegation of bad-faith, mental distress, oppression, fraud or insurance-related tort,” and would not “institute any class-action against defendant on account of the facts and issues involved in this loss and claim.”
After refusing to sign the foregoing agreement, plaintiffs sued, claiming that although their policy excluded coverage for losses caused or aggravated by earth movement, it implicitly provided coverage for losses caused by contractor negligence because negligence was not a specifically excluded peril under the policy. Plaintiffs also argued that defendant denied their claim before adequately investigating the damage to the structure, and that subsequent investigations were undertaken merely to confirm the original denial. In addition, plaintiffs asserted, defendant’s denial of coverage constituted a breach of the implied covenant of good faith and fair dealing and violated various provisions of the Insurance Code. Plaintiffs sought as relief (i) policy benefits, (ii) general damages for economic detriment and emotional distress, and (iii) punitive damages.
Defendant rested on the 12th day of trial, and the court granted a directed verdict for plaintiffs on the coverage issue. The court informed the parties it was following the decisions in Partridge, supra, 10 Cal.3d 94, and Sabella, supra, 59 Cal.2d 21, and that plaintiffs were covered under the policy because negligent construction, a covered risk, was a concurrent [401]*401proximate cause of the damage. Specifically, the trial court stated: “[The Supreme Court] told me in Sabella that negligent construction can be a proximate cause. They told me in Partridge there may be coverage whenever an insured risk constitutes simply a concurrent proximate cause of the injuries. [][] Now, to me that is crystal clear, putting those two causes together, that if negligent construction is a concurrent proximate cause of the loss, there is coverage.” The court continued, “The key witness for the defense, Mr. Nelson, conceded in his testimony, as I heard it and understood it, that the negligent construction was a cause of the room falling away. He did not use the word ‘proximate.’ He said a causative factor at one time. I don’t recall the exact language when he answered a question. In substance, that it was a cause on another occasion. As a matter of law, based upon the evidence, it was a proximate cause.”
The jury subsequently found defendant liable for $47,000 in policy benefits and general damages, and $1 million in punitive damages. The court denied defendant’s motions for judgment notwithstanding the verdict and for a new trial, and declined to issue a remittitur with respect to the punitive damages award. The court entered judgment in accordance with the verdict. Defendant appealed, and the Court of Appeal reversed the judgment in a divided opinion. Before reviewing the Court of Appeal holding, and in order to provide sufficient background information that will aid in the understanding of this case, we first discuss the development of multiple and concurrent causation insurance analyses, and the important distinction between property and liability policies.
II.
Discussion
A. Development of Multiple Causation Insurance Coverage Analyses
1. The efficient proximate cause standard
Our courts have long struggled to enunciate principles that determine whether coverage exists when excluded and covered perils interact to cause a loss. Initially, the courts attempted to reconcile section 530 (which provides for coverage when a peril insured against was the “proximate cause” of loss) with section 532 (which provides, that “If a peril is specifically excepted in a contract of insurance, and there is a loss which would not have occurred but for such peril, such loss is thereby excepted [from coverage] even though the immediate cause of the loss was a peril which was not excepted”).
[402]*402In our 1963 Sabella decision, supra, 59 Cal.2d 21, we faced a difficult property loss coverage question arising after a building contractor constructed a house on uncompacted fill and negligently installed a sewer line; negligent installation was a covered peril. Eventually, the sewer line ruptured causing water to saturate the ground surrounding the insureds’ house, resulting in subsidence, an excluded peril. The insureds brought a first party action against their insurer, seeking recovery for property loss under their homeowner’s property policy. (Id., at p. 26.)4 The trial court found the loss was not covered because subsidence was a specifically excluded peril under the policy. The insureds appealed this ruling and we reversed.
On its face, section 532 would have precluded coverage because the loss would not have occurred “but for” the excluded peril of subsidence. We recognized, however, that such a result would be absurd because it would deny coverage even though an insured peril “proximately” caused the loss simply because a subsequent, excepted peril was also part of the chain of causation. We reasoned that sections 530 and 532 were not intended to deny coverage for losses whenever “an excepted peril operated to any extent in the chain of causation so that the resulting harm would not have occurred ‘but for’ the excepted peril’s operation . . . .” (Sabella, supra, 59 Cal.2d at p. 33.) Rather, we explained that when section 532 is read along with section 530, the “but for” clause of section 532 necessarily refers to a “proximate cause” of the loss, and the “immediate cause” refers to the cause most immediate in time to the damage. (Id., at pp. 33-34.)
Thus, Sabella held that: “ ‘[I]n determining whether a loss is within an exception in a policy, where there is a concurrence of different causes, the efficient cause—the one that sets others in motion—is the cause to which the loss is to be attributed, though the other causes may follow it, and operate more immediately in producing the disaster.’ ” (59 Cal.2d at pp. 31-32, quoting from Couch on Insurance (1930) § 1466; Houser & Kent, Concurrent Causation in First-Party Insurance Claims: Consumers Cannot Afford Concurrent Causation (1986) Tort & Ins. L.J. 573, 575.)
Furthermore, in characterizing the “but for” clause of section 532 as referring to the efficient proximate cause of the loss, we impliedly recognized that coverage would not exist if the covered risk was simply a remote [403]*403cause of the loss, or if an excluded risk was the efficient proximate (meaning predominant) cause of the loss. On the other hand, the fact that an excluded risk contributed to the loss would not preclude coverage if such a risk was a remote cause of the loss.5
We relied heavily in Sabella, supra, 59 Cal.2d 21, on our earlier decision in Brooks v. Metropolitan Life Ins. Co. (1945) 27 Cal.2d 305 [163 P.2d 689], in which recovery was allowed on a homeowner’s policy insuring against death by accidental means. In Brooks, the insured, who was suffering from incurable cancer, an excluded peril, died in a fire. We held, “recovery may be had even though a diseased or infirm condition appears to actually contribute to cause the death if the accident sets in progress the chain of events leading directly to death, or if it is the prime or moving cause.” (Id., at pp. 309-310.) Brooks thus defined efficient proximate cause in the first party loss context as the “prime or moving cause.” (Ibid.)
The Court of Appeal here replaced the Sabella term “efficient proximate cause” with the term “moving cause.” Sabella defined “efficient proximate cause” alternatively as the “one that sets others in motion” (59 Cal.2d at p. 31), and as “the predominating or moving efficient cause.” (Id., at p. 32.) We use the term “efficient proximate cause” (meaning predominating cause) when referring to the Sabella analysis because we believe the phrase “moving cause” can be misconstrued to deny coverage erroneously, particularly when it is understood literally to mean the “triggering” cause. Indeed, we believe misinterpretation of the Sabella definition of “efficient proximate [404]*404cause” has added to the confusion in the courts and, in part, is responsible for the erroneous application of Partridge, supra, 10 Cal.3d 94, to first party property loss cases.
By relying on Brooks, supra, 27 Cal.2d 305 and construing sections 530 and 532, Sabella, supra, 59 Cal.2d 21, sets forth a workable rule of coverage that provides a fair result within the reasonable expectations of both the insured and the insurer whenever there exists a causal or dependent relationship between covered and excluded perils. In multiple cause cases, a proximate cause analysis, focusing on the efficient proximate cause, could be employed to determine whether or not the insured was covered for the loss under the property portion of the homeowner’s insurance policy. Indeed, for 10 years following Sabella, the Court of Appeal applied the efficient proximate cause analysis in resolving multiple-cause property-coverage questions under all-risk homeowner’s property policies. (See, e.g., Gillis v. Sun Ins. Office, Ltd. (1965) 238 Cal.App.2d 408, 415-420 [47 Cal.Rptr. 868, 25 A.L.R.3d 564] [coverage afforded under policy insuring loss by windstorm but excluding loss from water damage; wind, causing gangway to fall on and sink a dock, was deemed efficient proximate cause of loss]; Sauer v. General Ins. Co. (1964) 225 Cal.App.2d 275, 278-279 [37 Cal.Rptr. 303] [coverage afforded when water leaking from plumbing system (covered peril) was the efficient proximate cause of subsidence damage (excluded peril)].)
2. The doctrine of concurrent causation
In 1973, we were faced with a third party tort liability situation that presented a “novel question of insurance coverage” and did not fit the Sabella analysis because no single peril could be labeled the predominant cause of the loss. In Partridge, supra, 10 Cal.3d 94, the insured was covered under both an automobile liability policy and a homeowner’s liability policy with comprehensive personal liability coverage. (The latter liability policy excluded losses “arising out of the use” of a motor vehicle.) The insured, after filing the trigger mechanism of his pistol to create a “hair-trigger” action (such negligence was a covered risk under the homeowner’s property policy), hunted jackrabbits at night from his vehicle. As he drove over rough terrain while waving the gun in his hand (negligent driving was an excluded risk under homeowner’s liability policy), the gun fired and injured a passenger.
First party property coverage issues were not involved. The case concerned the personal liability of the insured who was sued by his injured passenger. Both policies were issued by the same insurer, which conceded the accident was covered under the automobile liability policy. The parties, [405]*405however, disputed whether coverage was also afforded under the homeowner’s liability policy, which excluded coverage for injuries “arising out of the use of motor vehicles.” As in Sabella, supra, 59 Cal.2d 21, we relied on Brooks, supra, 27 Cal.2d 305, at pages 309-310, to find coverage under the homeowner’s policy. In analyzing the liability coverage, we explicitly recognized, “the ‘efficient cause’ language [of Sabella] is not very helpful, for here both causes were independent of each other: the filing of the trigger did not ‘cause’ the careless driving, nor vice versa. Both, however, caused the injury .... If committed by separate individuals, both actors would be joint tortfeasors fully liable for the resulting injuries. Moreover, the fact that both acts were committed by a single person does not alter their nature as concurrent proximate causes. (Cf., Flournoy v. State of California (1969) 275 Cal.App.2d 806, 811 [80 Cal.Rptr. 485].)” (Partridge, supra, 10 Cal.3d at p. 104, fn. 10.) We concluded by stating, “Although there may be some question whether either of the two causes in the instant case can be properly characterized as the “ ‘prime,’ ‘moving’ or ‘efficient’ cause of the accident we believe that coverage under a liability insurance policy is equally available to an insured whenever an insured risk constitutes simply a concurrent proximate cause of the injuries.” (Id., at pp. 104-105, fns. omitted, second italics added.)
Because Partridge dealt with causation in the context of third party liability insurance, we did not address, nor did we contemplate, the application of our decision to the determination of coverage in the first party property insurance context. Indeed, Partridge asserted only that the “concurrent cause” standard was “consistent with Insurance Code sections 530 and 532, as authoritatively construed in Sabella v. Wisler. . . .” (Partridge, supra, 10 Cal.3d at p. 105, fn. 11.)
Furthermore, Partridge never considered in what manner concurrent causation could apply in the first party property insurance context. Rather, by recognizing in Partridge the “novel question” of liability coverage presented because two separate acts of negligence simultaneously joined together to cause an injury, we also impliedly recognized the limited scope of our holding. We did not extend our holding to first party property insurance cases. Accordingly, we should not apply the decision to such cases merely because it appears to simplify the coverage analysis.
B. The Distinction Between Liability and Property Insurance
As we will demonstrate, the decision of the Court of Appeal, in applying the Partridge concurrent causation analysis to property damage cases, like other recent insurance cases involving all-risk homeowner’s policies, failed to differentiate between property loss coverage under a first party policy and [406]*406tort liability coverage under a third party policy of insurance. First- and third party coverage is today typically provided in a single policy, and under both types of coverage, once the insured shows that an event falls within the scope of basic coverage under the policy, the burden is on the insurer to prove a claim is specifically excluded. (See Clemmer v. Hartford Ins. (1978) 22 Cal.3d 865, 880 [151 Cal.Rptr. 285, 587 P.2d 1098]; Royal Globe Ins. Co. v. Whitaker (1986) 181 Cal.App.3d 532, 537 [226 Cal.Rptr. 435]; Strubble v. United Services Auto. Assn. (1973) 35 Cal.App.3d 498, 504 [110 Cal.Rptr. 828].) Moreover, exclusionary clauses are interpreted narrowly, whereas clauses identifying coverage are interpreted broadly. (See Reserve Ins. Co. v. Pisciotta (1982) 30 Cal.3d 800, 808 [180 Cal.Rptr. 628, 640 P.2d 764].)
The scope of coverage and the operation of the exclusion clauses, however, are different in the separate policy portions and should be treated as such. As one commentator has recently stated: “Liability and corresponding coverage under a third party insurance policy must be carefully distinguished from the coverage analysis applied in a first party property contract. Property insurance, unlike liability insurance, is unconcerned with establishing negligence or otherwise assessing tort liability.” (Bragg, Concurrent Causation and the Art of Policy Drafting: New Perils for Property Insurers (1985) 20 Forum 385, 386.)
For these reasons it is important to separate the causation analysis necessary in a first party property loss case from that which must be undertaken in a third party tort liability case. The following quotation summarizes the distinction that must be drawn: “Property insurance . . . is an agreement, a contract, in which the insurer agrees to indemnify the insured in the event that the insured property suffers a covered loss. Coverage, in turn, is commonly provided by reference to causation, e.g., ‘loss caused by . . .’ certain enumerated perils. [^|] The term ‘perils’ in traditional property insurance parlance refers to fortuitous, active, physical forces such as lightning, wind, and explosion, which bring about the loss. Thus, the ‘cause' of loss in the context of a property insurance contract is totally different from that in a liability policy. This distinction is critical to the resolution of losses involving multiple causes, [if] Frequently property losses occur which involve more than one peril that might be considered legally significant. If one of the causes (perils) arguably falls within the coverage grant—commonly either because it is specifically insured (as in a named peril policy) or not specifically excepted or excluded (as in an “all risks” policy)—disputes over coverage can arise. The task becomes one of identifying the most important cause of the loss and attributing the loss to that cause.” (Bragg, supra, 20 Forum at pp. 386-387, italics added.)
[407]*407On the other hand, the right to coverage in the third party liability insurance context draws on traditional tort concepts of fault, proximate cause and duty. This liability analysis differs substantially from the coverage analysis in the property insurance context, which draws on the relationship between perils that are either covered or excluded in the contract. In liability insurance, by insuring for personal liability, and agreeing to cover the insured for his own negligence, the insurer agrees to cover the insured for a broader spectrum of risks. In order to further demonstrate the differences between property loss and liability coverage, we compare two sections of a typical homeowner’s policy—the all-risk property loss coverage section of the policy in this case and the personal liability section at issue in Partridge.
Each policy section is governed by separate exclusions. For example, the all-risk first party property loss coverage section in this case provides for coverage against “all risk of physical loss” except: losses “caused by . . . settling cracking, shrinkage, bulging or expansion of pavements, patios, foundations, walls, floors, roofs or ceilings . . . .”
In comparison, in Partridge, “The coverage clause of the ‘Personal Liability’ section of the ‘Homeowner’s Policy’ provides in relevant part: ‘This Company agrees to pay on behalf of the Insured all sums which the Insured shall become legally obligated to pay as damages because of bodily injury or property damage, to which this insurance applies, caused by an occurrence.’ . . .” (10 Cal.3d at p. 99, fn. 5.) Moreover, “[t]he applicable exclusionary clause reads: ‘This policy does not apply: 1. Under Coverage E—Personal Liability . . . (a) To Bodily Injury or Property Damage Arising Out of the Ownership, Maintenance, Operation, Use, Loading or Unloading of: . . . (2) Any Motor Vehicle Owned or Operated By, or Rented or Loaned to, any Insured ....’” (10 Cal.3d at p. 99, fn. 6.)
As the two provisions cited above illustrate, under the all-risk first party property policy, because generally “all risk of physical loss” is covered, the exclusions become the limitation on loss coverage. Under the liability por[408]*408tion of the policy, on the other hand, the focus is, at least initially, on the insured’s legal obligation to pay for injury or damage arising out of an “occurrence.”
In the property insurance context, the insurer and the insured can tailor the policy according to the selection of insured and excluded risks and, in the process, determine the corresponding premium to meet the economic needs of the insured. On the other hand, if the insurer is expected to cover claims that are outside the scope of the first party property loss policy, an “all risk” policy would become an “all loss” policy. (Friedman, Concurrent Causation: The Coverage Trap (1985) 86 Best’s Rev.: Prop./Casualty 50, 58.) In most instances, the insured can point to some arguably covered contributing factor. As we shall discuss, if the rule in Partridge, supra, 10 Cal.3d 94, were extended to first party cases, the presence of such a cause, no matter how minor, would give rise to coverage.
Finally, as we explain, the reasonable expectations of the insurer and the insured in the first party property loss portion of a homeowner’s policy—as manifested in the distribution of risks, the proportionate premiums charged and the coverage for all risks except those specifically excluded—cannot reasonably include an expectation of coverage in property loss cases in which the efficient proximate cause of the loss is an activity expressly excluded under the policy. Indeed, if we were to approve of the trial court’s directed verdict, we would be requiring ordinary insureds to bear the expense of increased premiums necessitated by the erroneous expansion of their insurers’ potential liabilities. (See Hartford Fire Ins. Co. v. Superior Court (1983) 142 Cal.App.3d 406, 417 [191 Cal.Rptr. 37, 39 A.L.R.4th 189].)
C. The Court of Appeal Holding
In reversing the directed verdict, the Court of Appeal rejected defendant’s argument that although third party negligence is not specifically excluded in an all-risk policy, it is not a covered peril because it is technically not considered a “risk of physical loss” within the policy terms. We agree and find defendant’s claim is not supported by authority. (See, e.g., Sabella, supra, 59 Cal.2d 21, 31.)7
[409]*409Next, the court recognized that recent first party property loss cases (discussed below) have forsaken the efficient proximate cause analysis developed in Sabella, supra, 59 Cal.2d 21, and have looked instead to the holding in Partridge, supra, 10 Cal.3d 94, to allow coverage for property damage simply because a nonexcluded risk is an independent proximate cause of the loss. Although the Court of Appeal refused to confine the application of Partridge to liability cases, the court did attempt to limit the scope of Partridge's concurrent causation standard in both liability and property cases by requiring the included and excluded perils to be independent of origin and independent in operation. Specifically, the Court of Appeal interpreted Partridge, supra, 10 Cal.3d 94, as holding that “because each act could have caused the loss regardless of the existence of the other act [citation], they were independent of each other.” According to the Court of Appeal: “Neither [cause in Partridge] can be said to have necessarily acted upon a condition created by the other, or to have propelled the other, or to have brought to fruition the potential for damage inherent in the other.” The Court of Appeal believed the “Partridge court would not have found the covered risk to be ‘independent’ if that risk (negligence) existed only in relation to, [or was dependent on] an excluded risk.” Thus, the Court of Appeal determined that in order for coverage to be found under Partridge, the concurrent event alone must have been a “sufficient condition” of the loss—i.e., capable of producing damage itself.8
In its effort to interpret and apply the Partridge decision in the present case, however, the court followed the lead of other Court of Appeal decisions and assumed, without discussion, that Partridge should apply to first party property insurance policies. In so doing, the court relied on both Partridge, and Sabella, supra, 59 Cal.2d 21, and announced the following two-tiered rule for determining coverage: “[W]hether the covered risk and excluded risk are causes in fact should be a court’s threshold inquiry in cases such as this. If (i) they both are causes in fact and if the two risks are independent of each other, Partridge analysis is triggered: the insured is covered if the covered risk was a concurring proximate cause of the loss. If (ii) the two risks are dependent on each other, Sabella analysis is triggered: the insured is covered only if the covered risk was the moving cause of the loss.” The Court of Appeal determined that the jury in the present case [410]*410must decide the coverage issue because it was not clear, as a matter of law, whether the two risks were dependent on or independent of each other.
The Court of Appeal’s formulation was an attempt to provide trial courts with an analytical framework from which to determine property insurance coverage when a loss is arguably caused by concurrent causes. The court was on the right track in attempting to reconcile first and third party cases that use the Partridge analysis in determining coverage. We believe, however, as set forth above, that the court erroneously failed to limit at the threshold the application of Partridge to the third party liability context.9
D. The Misapplication of Partridge in the First Party Property Insurance Context
In Safeco Ins. Co. of America v. Guyton, supra, 692 F.2d 551, 553, the insurer sought declaratory relief against several insureds after flooding damaged their homes. The trial court rejected the policyholders’ claim that the water district’s negligence in failing to provide adequate flood control facilities (a covered risk) caused the damage, and held under Sabella there was no coverage because the excluded peril—flooding—was the efficient proximate cause of the property damage. The Ninth Circuit Court of Appeals reversed on the basis that the insureds should be allowed to seek coverage under Partridge, supra, 10 Cal.3d 94. The court determined that in order for Partridge to apply, the covered peril must have existed independently of the excluded peril. The court reasoned, however, that the requirement was met because two independently created conditions interacted to cause the flood damage. Indeed, Guyton specifically stated that “the twin causes in Partridge were independent only in the sense that each cause had an independent origin, not that they did not interact with one another to cause the loss.” (692 F.2d at p. 555.)
As amici curiae for defendant point out, however, Guyton, supra, 692 F.2d 551, was actually “a classic case of dependent causation” requiring use [411]*411of a Sabella analysis. Because the damage caused by the defective flood control system was necessarily dependent on flooding, the Ninth Circuit misapplied Partridge to find coverage. It should have looked to Sabella for resolution to determine whether the defectively maintained flood control system was the efficient proximate cause of the property losses even though the flood was the “immediate” cause of the losses, or whether the trial court correctly denied coverage in determining that flooding was the efficient proximate cause of the loss. The Ninth Circuit specifically ignored the effect of the express policy exclusion contained in the Safeco policy that excluded loss “caused by, resulting from, contributed to or aggravated by any of the following: [H] a. Flood, surface water. . . .” (Guyton, supra, 692 F.2d at pp. 552-553.)
In Premier Ins. Co. v. Welch, supra, 140 Cal.App.3d 720, 728, a third party’s negligence (a covered peril) damaged a subdrain. After a heavy rain, the house slid from its foundation. Water damage was an excluded peril. The trial court relied on Sabella, supra, 59 Cal.2d 21, and determined that coverage should be denied because the efficient proximate cause of the loss was the rainfall. The same Court of Appeal division that decided the present case reversed and awarded coverage, concluding under a Sabella analysis that the efficient proximate cause of the loss was the damaged subdrain. Notwithstanding this conclusion, the court continued by stating that the insured would be covered in any event under Partridge, supra, 10 Cal.3d 94, because “as a matter of law . . . the damage to the drain was a concurrent proximate cause of the loss.” (Premier, supra, 140 Cal.App.3d at p. 728.)
In its opinion in the present case, the Court of Appeal correctly acknowledged that its earlier interpretation (in Premier) of Partridge’s independence requirement was wrong. Reviewing the facts of Premier the court stated that the property loss caused by the negligently damaged subdrain was dependent on the existence of rainfall, but that nonetheless, the slide would not have occurred if the drain had not been severely damaged. Thus, the Court of Appeal recognized, “Premier was simply a Sabella situation.” As the court admitted, it had reached a satisfactory result under Sabella, supra, 59 Cal.2d 21; the Partridge analysis was simply unnecessary in Premier.10
[412]*412E. The Trial Court Erred in Directing the Verdict
Finally, because we believe the Partridge analysis should be limited to third party liability cases, we cannot sustain the directed verdict in this case. Plaintiffs argue the directed verdict was proper because there was no evidence of sufficient substantiality to support a determination other than the following: the included risk of negligent construction and the excluded risk of earth movement in the form of soil creep, if any, were independent, concurrent proximate causes of the loss plaintiffs sustained to their property.
We disagree. The record does not support either plaintiffs’ contention or the hyperbolic characterization of the expert testimony in Justice Mosk’s dissent. The record discloses that the experts were in marked disagreement as to the causes of plaintiffs’ damages. For example, Mr. Nelson, the defense expert, testified: “There was settlement and creep. The room addition moved away from the house and was caused by—probably caused by settlement and creep.” Plaintiffs’ expert, Mr. Hillebrandt testified to the contrary. To the question, “Was what happened to the room addition in August 1978 related to any respect to soil creep, in your opinion?” Hillebrandt replied, “In my opinion, it didn’t. And the reason is because the actual breaking—the footing of the support beam was actually on a level surface. It was on a level fill surface that had been placed behind the rock wall, and the whole hillside may have been creeping, but this was just one part of it, and it was a level surface.” At another point during the trial, however, Hillebrandt testified that the damage to the deck was caused by the following: “In this case, the footing [of the retaining wall] was not deep enough, so consequently, when the creep forces at the surface caused this downhill movement, the footing just went along for the ride.”
This case presents a classic Sabella situation. Coverage should be determined by a jury under an efficient proximate cause analysis. Accordingly, bearing in mind the facts here, we conclude the question of causation is for the jury to decide. 11 If the earth movement was the efficient proximate cause of the loss, then coverage would be denied under Sabella, supra, 59 Cal.2d 21. On the other hand, if negligence was the efficient proximate cause of the [413]*413loss, then coverage exists under Sabella. These issues were jury questions because sufficient evidence was introduced to support both possibilities.
The judgment of the Court of Appeal is affirmed with directions to remand the cause to the trial court for further proceedings consistent with the opinion of this court.
Panelli, J., Eagleson, J., and Arguelles, J.,* concurred.
Retired Associate Justice of the Supreme Court sitting under assignment by the Chairperson of the Judicial Council.