I agree with the plurality that the insurer did not waive its *157“multiple occurrence” and “pro rata allocation” arguments by failing to timely assert these issues in its disclaimer letters. I also concur with its conclusion that pro rata allocation of the loss across all implicated policies is appropriate.1 But because I do not believe that application of the “unfortunate event” test results in a finding that the continuous course of sexual abuse of a single child by the same negligently hired and supervised priest amounted to multiple occurrences, I respectfully dissent from that part of the plurality decision.
I
Under the National Union policy, the term “occurrence” is defined as “an accident including continuous or repeated exposure to substantially the same general harmful conditions.” This definition is similar to the clause we interpreted in Appalachian Ins. Co. v General Elec. Co. (8 NY3d 162 [2007]), where we applied the unfortunate event test to hold that General Electric could not aggregate, on an annual basis, asbestos-related personal injury claims brought by numerous individuals who were exposed to asbestos-containing steam turbines installed at more than 22,000 sites throughout the United States.
We made clear in Appalachian that resolving the occurrence question is a two-part process. First, the court must identify the incidents or occasions giving rise to injury or loss. Second, it must apply the unfortunate event test to determine whether they constitute one occurrence. Under the second inquiry, the court considers “whether there is a close temporal and spatial relationship between the incidents . . . , and whether the incidents can be viewed as part of the same causal continuum, *158without intervening agents or factors” (id. at 171-172). In Appalachian, we explained that “the incident that gave rise to liability was each individual plaintiff’s continuous or repeated exposure to asbestos” (id. at 173 [internal quotation marks omitted]). Applying the unfortunate event factors, we concluded there were numerous occurrences because, even assuming the causal continuum element was met, “the incidents share few, if any, commonalities, differing in terms of when and where exposure occurred, whether the exposure was prolonged and for how long, and whether one or more GE turbine sites was involved” (id. at 174). This case is a far cry from Appalachian.
Here, plaintiffs in the underlying personal injury action sought recovery from the Roman Catholic Diocese of Brooklyn based on its alleged negligent hiring and supervision of a priest who, over a span of more than six years, repeatedly engaged in a course of sexual abuse of a particular child. Just as the incident or occasion giving rise to injury or loss in Appalachian was the repeated or continuous exposure of an individual plaintiff to asbestos, here the incident or occasion was the repeated or continuous exposure of the child to the same negligently hired and supervised priest.
Although the course of sexual conduct allegedly engaged in by the priest constitutes intentionally tortious (indeed criminal) behavior, the Diocese was sued for its own allegedly negligent acts: inappropriate hiring, retention and supervision of the priest. Viewed from the perspective of the Diocese—the entity that purchased insurance coverage—the incident giving rise to liability was the child’s repeated molestation by the same priest, which was a consequence of her exposure to the same risk of harm created by the Diocese (i.e., an allegedly negligently hired and improperly supervised employee). As noted in Appalachian, the fact that the exposure was repeated and continued for several years “does not make it any less the operative incident or occasion giving rise to liability” (8 NY3d at 174). In reaching this conclusion, I do not conflate the cause of the incident with the incident itself as the plurality suggests. In fact, I apply the unfortunate event test to the same “incident or occasion” as the plurality—the priest’s sexually abusive contact with the child. Our differences arise from disparate views concerning the policy language and the application of the unfortunate event factors.
Contrary to the conclusion reached by the plurality, by defining a single occurrence as encompassing “continuous or *159repeated exposure to substantially the same general harmful conditions,” the insurance contract included language in the policy evincing an intent to aggregate certain bodily injury events that might otherwise be viewed as distinct incidents. The purpose of the unfortunate event test is to determine when such aggregation is appropriate.
Applying the unfortunate event factors, I conclude that there was one occurrence. First, the claim for which the Diocese seeks coverage arose from the bodily injury of one party. None of the possible tests discussed in Appalachian (the proximate-cause approach, the one-accident-per-person approach, or the unfortunate event test) would have resulted in separating into multiple occurrences a claim by one complainant suing based on exposure to the same injury-producing condition (absent language restricting the duration of an occurrence, which I discuss below). Certainly, application of the test we adopted in Appalachian does not require that result.
Although the abuse incidents continued for more than six years, there were no substantial periods of intervening time when there was no abuse. Rather, according to the submissions in the personal injury action, for several of the years encompassing instances of abuse the child’s mother worked as a cook at the rectory at least five days a week and sometimes on weekends; she would pick the child up at school and bring her to work every day, where the child would remain until the mother went home, sometimes late at night. The infant plaintiff indicated that the priest’s inappropriate sexual contact occurred virtually every day that the child was at the rectory and sometimes multiple times in the same day. Thus, there is no question that the sexual abuse incidents were sufficiently frequent and connected (i.e., temporally close to one another) to meet the requirements of the unfortunate event test.
Nor is there a “spatial” disconnect, to the extent that factor can be applied here. Most of the abuse incidents occurred at the rectory—all arose from the child’s exposure to the same priest, which came about as a result of his employment with the Diocese where her mother worked. And, at least insofar as the Diocese’s liability is concerned, the incidents had a common cause—the Diocese’s negligent hiring, retention and supervision of the priest. It is true, as the plurality notes, that the priest engaged in numerous acts of sexual abuse—the Diocese’s alleged negligence was certainly not the sole cause of the child’s *160injuries. But this did not disrupt the causal link between the Diocese and the incidents because the Diocese was held liable for a continuous course of conduct—negligent hiring, supervision and retention of the priest—that created the occasion for the abuse. Just because the Diocese’s negligence was not the sole cause of the injuries does not mean that there was a break in the causal chain between its acts or omissions and the incidents underlying the injured plaintiffs loss. I therefore conclude that the sexual abuse constituted one occurrence under the unfortunate event test.
Purporting to apply that test, the plurality instead holds that there were multiple occurrences—although the decision is somewhat unclear as to whether there were multiple occurrences per policy or whether it is aggregating abuse incidents within each policy year, resulting in one occurrence per policy. The difference between these two conclusions can be significant (especially when there is an excess insurer that has also provided coverage on a “per occurrence” basis, as is the case here). Reading between the lines, because it emphasizes the disparate nature of each sexual abuse incident, and there were numerous assaults during each policy period, it appears that the plurality concludes that there were multiple occurrences per policy. If the priest sexually abused the child 20 times in a given year (and this is an extremely low estimate given her allegations), then under the plurality’s one-occurrence-per-abuse-incident analysis the Diocese would be required to pay 20 $250,000 self-insured retentions for that period—even though its liability emanated from a single lawsuit involving one injured party who was exposed to the same repeated, virtually-continuous risk of harm.
I cannot imagine that this was what the parties had in mind when the coverage was purchased. Certainly, the policy contemplated that the Diocese might seek coverage in any given policy period for more than one occurrence—in the course of a year, the Diocese might be liable for several different injury-producing events (e.g., a parishioner might fall on the church steps; a child might be injured at a Diocesan school playground; a tree located on church property might fall on a visitor’s vehicle resulting in property damage). But what the parties would not have anticipated was that a single lawsuit brought by one injured plaintiff who suffered damages as a result of the same harmful condition would be treated as multiple occurrences within each policy period. While each sexual abuse incident may well have constituted a separate “bodily injury,” the parties *161agreed that the self-insured retention would be triggered by an “occurrence.” Since the policy defines an occurrence as the “continuous or repeated exposure to substantially the same general harmful conditions,” and the temporal, spatial and causal analysis indicates that the abuse was part of one unfortunate event, there was only one occurrence.
II
That being said, my conclusion that the course of sexual conduct involving the infant plaintiff involved one occurrence under the unfortunate event test does not necessarily resolve the question of how many self-insured retentions the Diocese was required to pay when it sought coverage under more than one policy. Parties to insurance contracts are free to fashion self-insured retention provisions that require that a retention be paid every time a policy’s coverage is triggered, regardless of whether this is based on the same injury-producing condition. The question is whether that occurred here.
As the plurality explains, each of the National Union policies provide coverage for bodily injury only if it is “caused by an occurrence” and “occurs during the policy period.” Here, since there was bodily injury (sexual abuse incidents) caused by an occurrence that spanned multiple policy periods, there is no question that multiple coverage provisions were triggered. However, in these policies, the Diocese’s obligation to pay a self-insured retention does not hinge on whether there has been “bodily injury” within a policy period—rather, the policies require the payment of one self-insured retention per occurrence.
In many insurance policies, an occurrence cannot span more than one policy year because the definition of “occurrence” contains language restricting its temporal scope. This was true in Appalachian where the policies defined “occurrence” as “an accident, event, happening or continuous or repeated exposure to conditions which unintentionally results in injury or damages during the policy period” (8 NY3d at 172 [emphasis added]). But in this case, though the parties used temporally limiting language to define coverage for “bodily injury,” the definition of “occurrence” contains no similar language restricting its duration.2 By agreeing that one self-insured retention would be owed *162per “occurrence” and failing to include language restricting an occurrence to a single policy term, the insurance agreement required the Diocese to pay only one self-insured retention for an occurrence even if it spanned multiple policy years—triggering multiple policies—as occurred here. I would therefore modify the decision of the Appellate Division by issuing a declaration to that effect.
The cases from other jurisdictions cited by National Union are not to the contrary. Most involved insurance policies in which the definition of “occurrence” included a clause comparable to the restrictive “during the contract term” language found in Appalachian (and present in many if not most occurrence-based policies)—which is inexplicably absent here. This was true in Interstate Fire & Cas. Co. v Archdiocese of Portland in Or. (35 F3d 1325 [9th Cir 1994]) where an excess insurer brought suit to resolve a dispute concerning the primary coverage available for a settlement involving the repeated sexual abuse of a single child that continued over four policy periods. The excess insurer contended that each act of molestation was a separate occurrence, meaning that, for each incident of sexual abuse, the Diocese would be required to pay a self-insured retention and the primary insurer would have to exhaust its per-occurrence limit. The Diocese maintained that all four years of molestation involved one occurrence, implicating one self-insured retention and one per-occurrence primary policy limit. The 9th Circuit rejected both arguments based on the language of the policy, which defined occurrence as “an accident or a happening or event or a continuous or repeated exposure to conditions which unexpectedly and unintentionally results in personal injury . . . during the policy period” (id. at 1329 [emphasis added]).
*163Similar to my analysis, the 9th Circuit reasoned: “it is the repeated ‘exposure’ of the [child] to the negligently supervised priest, resulting in injury, that provides the basis for indemnification” and, as such, “it is ‘exposure’ to such conditions . . . that constitutes the occurrence” (id.) It went on to explain:
“Although the definition of occurrence provides that multiple exposures stemming from the same general conditions can constitute a single occurrence, the assuring clause makes it clear that this is true only of multiple exposures occurring during the period of insurance” (id. at 1329-1330).
Because the language in the pertinent policies precluded an occurrence from spanning more than one policy period, the court determined there was one occurrence per policy year. The plurality cites the case (plurality op 152-153 n 5) but misunderstands it, suggesting that the 9th Circuit refused to group together individual acts of sexual abuse when, in fact, the court aggregated all acts occurring within a single policy year as one occurrence and found four total occurrences only because of the policy’s durational restriction on the scope of an occurrence. The Interstate Fire analysis is sound but this case is distinguishable because the policies under review fail to restrict the duration of an occurrence.
Since I concur with the plurality’s ultimate conclusion on pro rata allocation, there is no reason for me to discuss allocation— other than to clarify the difference between the occurrence and allocation issues. The distinction I draw between this case and Interstate Fire (which also did not resolve an allocation issue) is not related to our loss allocation rule but pertains to the occurrence issue and is based on differences in the policy definitions of an “occurrence.” Although there are elements of the occurrence and allocation questions that overlap (e.g., both consider causation), whether pro rata allocation is appropriate involves a distinct inquiry that does not necessarily depend on the number of occurrences involved.3 When a policy contains language restricting the duration of an occurrence to a single policy period, a continuous harm can constitute multiple occurrences *164(one per policy period) but still warrant pro rata allocation of the loss across all implicated policies, as appears to have been the case in Consolidated Edison Co. (98 NY2d 208 [2002], supra [pro rata allocation deemed appropriate in case where occurrence definition contained typical “during the policy period” restriction]). And certainly where, as here, there is no language restricting the duration of an occurrence, a continuous harm that causes losses across multiple policy periods can constitute a single occurrence and also present an appropriate candidate for pro rata allocation.
The bottom line is that, whether based on the policy language or some other analysis, most negligent hiring and supervision cases arising from repeated acts of child sexual abuse generally fall into one of two categories. In cases where the abuse did not span more than one policy period, courts have held that there was one occurrence per injured plaintiff (State Farm Fire & Cas. Co. v Elizabeth N., 9 Cal App 4th 1232, 12 Cal Rptr 2d 327 [1992] [where several children were abused by husband of child care provider multiple times over a one-month period, insured’s liability to each child constituted a single occurrence]; S.F. v West Am. Ins. Co., 250 Va 461, 463 SE2d 450 [1995] [where an employee sexually assaulted more than one person in separate incidents, each exposure to a new victim was a new occurrence]; see generally H.E. Butt Grocery Co. v National Union Fire Ins. Co. of Pittsburgh, Pa., 150 F3d 526 [5th Cir 1998] [did not involve a course of sexual abuse but, in case where employee assaulted two children on two separate occasions one week apart, there was one occurrence per injured plaintiff]). In cases where the definition of occurrence included durationally-restrictive language and the abuse spanned more than one policy year, courts have concluded that there was one occurrence per injured plaintiff per policy year (see Society of R.C. Church of Diocese of Lafayette & Lake Charles, Inc. v Interstate Fire & Cas. Co., 26 F3d 1359 [5th Cir 1994] [where two priests at one Diocese abused 31 children over seven years, there was one occurrence per child per policy year]; Roman Catholic Diocese of Joliet, Inc. v Interstate Fire Ins. Co., 292 Ill App 3d 447, 685 NE2d 932 [1997] [relationship between priest and minor spanning two policy years constituted one occurrence per policy period]). As *165far as I can detect, none of the cases involving a course of sexual conduct against a single child have employed the analysis adopted by the plurality here, which suggests that each act of sexual abuse involving the same victim constitutes a separate occurrence. Hence, because I find the plurality approach on the occurrence issue to be inconsistent with the policy language and the pertinent precedent, I respectfully dissent from that part of the decision.
Judges Read and Pigott concur with Judge Rivera; Judge Smith concurs in result in an opinion; Judge Graffeo concurs in part and dissents in part in an opinion; Chief Judge Lippman taking no part.
Order affirmed, with costs, and certified question answered in the affirmative.
. The pro rata allocation and multiple occurrence issues are distinct. Under the allocation issue, the question is whether the insurer is entitled to demand that the loss be apportioned across all seven policies in effect when the abuse occurred, rather than covered by only one or two at the election of the insured. Resolution of the allocation issue is significant, no matter how the Court decides the occurrence question, because the Diocese indicates that several of the primary policies contained a sexual abuse exclusion that would preclude coverage for losses in those policy years. If the Diocese is correct about the scope of the policy exclusions (an issue not before us), since we have concluded that the claim must be allocated across all seven policy periods, it appears that the Diocese may recover only the pro rata share attributable to the first two policies which allegedly had narrower sexual abuse exclusions and therefore may cover the loss. Resolution of the occurrence issue involves a different question: how many self-insured retentions will the Diocese have to pay to gain access to the pro rata coverage allegedly available under the policies that had the narrower exclusions.
. In this regard, Judge Smith’s concurrence fails to recognize the relevant language in the insurance contract. He suggests that there is no distinction *162between a covered “injury” and an “occurrence” (Smith, J. concurring op at 156)—but in this policy the terms “bodily injury” and “occurrence” have separate functions and different definitions. It is only an occurrence that triggers the obligation to pay a self-insured retention, with one self-insured retention owed for each occurrence. Although he agrees that there was one occurrence under the unfortunate event test, Judge Smith does not explain why he nonetheless concludes that there were multiple occurrences—one per policy period—for purposes of payment of the self-insured retention. His reliance on Consolidated Edison Co. of N.Y. v Allstate Ins. Co. (98 NY2d 208 [2002]) is misplaced because in that case we did not determine either the number of occurrences or the scope of a self-insured retention requirement, nor indicate that “a new retention applies to the ‘injury’ suffered each year” (Smith, J. concurring op at 156); the case addresses a different question—how a loss arising from a continuing harm should be allocated.
*161 (n. cont’d)
. Absent policy language to the contrary, pro rata allocation is appropriate if (1) the loss arose as a result of a continuous harm spanning multiple policy periods; and (2) it is difficult if not impossible to discern the extent of injuries attributable to any one policy period (Consolidated Edison, 98 NY2d at 221, supra). I agree with Judge Smith that it is hard to reconcile the plurality’s view that the sexual abuse incidents are too causally attenuated to *164be aggregated as a single occurrence (even within the same policy period) with its subsequent treatment of the pro rata allocation issue, which depends on the conclusion that the entire loss stems from the same continuous harm spanning multiple policy periods.
*163 (n. cont’d)