Sixth Circuit Clarifies “Number of Occurrences” Analysis in J.M. Smucker Salmonella Coverage Ruling

When ACE American Insurance Company denied The J.M. Smucker Company’s claim following a 2022 salmonella-linked peanut butter recall, the dispute was never about whether coverage existed. It was about how many times Smucker would have to satisfy a $250,000 retained limit before that coverage attached. ACE’s answer was 225 times. On July 1, 2026, the United States Court of Appeals for the Sixth Circuit rejected that position, holding that the contamination behind the recall, rather than each of the thousands of consumer claims that followed, was a single “occurrence” under Smucker’s commercial general liability policies.¹ The decision, J.M. Smucker Co. v. ACE American Insurance Co., is a useful vehicle for revisiting one of the more consequential and recurring questions in liability insurance: how courts count “occurrences” when a single defective product or process injures many people over time.

Why the Number of Occurrences Question Matters

Most CGL and product-liability policies apply per-occurrence limits, and many carry a deductible or self-insured retention that also runs on a per-occurrence basis. Ordinarily this framework is straightforward: one accident, one injured party, one occurrence. The analysis becomes far more consequential, and far more contested, when a single product defect, contamination event, or design flaw produces injuries to hundreds or thousands of claimants over an extended period. If each exposure or each claimant counts as a separate occurrence, an insurer’s practical exposure can shrink dramatically, because the insured must exhaust a separate retention before coverage attaches to each claim, and per-occurrence limits can be reached, and aggregate limits exhausted, far more quickly across many “occurrences.” If the event instead counts as a single occurrence, the insured pays one retention, and the full per-occurrence limit remains available to respond to the entire mass of claims. In Smucker, the stakes were stark: under ACE’s 225-occurrence theory, Smucker would have had to satisfy $56.25 million in retained limits under each of its two applicable policy years, $112.5 million combined, before ACE owed anything, versus a single $250,000 retained limit under Smucker’s single-occurrence theory.²

The National Split: Cause Test Versus Effect Test

Courts nationally have divided for decades over how to answer the number-of-occurrences question, generally along two competing frameworks. Under the “cause” test, courts look to the underlying cause of the injuries, the act or omission by the insured that set the harm in motion, rather than to the number of resulting injuries or claims. If a single, uninterrupted, and continuing cause produced all of the damage, there is one occurrence, regardless of how many people were ultimately affected. Under the “effect” test, by contrast, courts count occurrences by looking to the number of distinct injuries or claimants, so that each person harmed, or each discrete event in which harm occurred, can constitute a separate occurrence even where a single defect or condition is the common thread. Ohio, whose law governed the Smucker policies, applies the cause test: the number of occurrences turns on the cause of the damage or injury, not on the number of individual claims.³ The Ninth Circuit applied the cause test decades ago in Chemstar, Inc. v. Liberty Mutual Insurance Co., a diversity case predicting California law and following its own earlier precedent, holding that a supplier’s failure to warn end users that its high-periclase lime was unsuitable for interior use, a failure traceable to inadequate quality control, was a single occurrence even though the resulting pitting damaged twenty-eight homes.⁴ The Fifth Circuit applied the same cause-based framework in Evanston Insurance Co. v. Mid-Continent Casualty Co., treating a runaway truck’s successive collisions at a Houston toll plaza as a single “accident” because the driver’s ongoing negligence was the single proximate, uninterrupted, and continuing cause of the resulting damage.⁵

The Smucker Dispute

Smucker purchased consecutive year-long CGL policies from ACE, one covering 2021 and a second covering 2022, each carrying a $250,000 retained limit per occurrence and defining “occurrence” as “an accident, including continuous or repeated exposure to substantially the same general harmful conditions.” In 2022, Smucker recalled peanut butter products manufactured at its Lexington, Kentucky facility after learning of potential salmonella contamination. Thousands of consumers filed claims alleging bodily injury and property damage. When Smucker sought coverage, ACE took the position that each claimant’s individual exposure to the contaminated product was its own occurrence, then invoked a “Lot Endorsement” in the policies to aggregate those thousands of individual occurrences into 225 occurrences, one for each roughly 24-hour production “lot” implicated in the recall.⁶ Smucker sued for breach of contract and a declaration that a single occurrence, and a single retained limit, applied. The United States District Court for the Northern District of Ohio agreed with Smucker on cross-motions for summary judgment, found the Lot Endorsement ambiguous, and, on ACE’s motion, certified the order for interlocutory appeal and stayed the underlying case pending review.⁷

The Policy Definition Pointed to a Single Accident

Reviewing the summary judgment ruling de novo, the Sixth Circuit, in an opinion by Judge Siler joined by Judges Moore and Bloomekatz, worked through the question in three analytical steps. The court first parsed the policies’ own definition of “occurrence.” Because the policies did not define “accident,” the court gave the word its ordinary meaning: something that happens by chance, without deliberate cause.⁸ Critically, under Ohio law, courts assess the accident from the insured’s own perspective, asking what the insured did unintentionally that exposed it to liability, rather than what happened to each person the insured’s conduct eventually affected.⁹ Viewed that way, Smucker’s only identifiable accident was the unintentional production of contaminated peanut butter; each consumer’s later decision to eat the product was neither an accident nor Smucker’s own conduct, and so could not multiply the number of occurrences. The court found support in its own prior decision in Scott Fetzer Co. v. Zurich American Insurance Co., where the operative “accident” for insurance purposes was the negligent hiring and supervision of an independent dealer of the insured’s products who sexually assaulted three women, not the three individual assaults themselves.¹⁰ The “continuous or repeated exposure to substantially the same general harmful conditions” language in Smucker’s policies reinforced the same conclusion: the salmonella contamination was a single, continuous harmful condition, not a series of discrete conditions capable of generating separate occurrences.

Ohio’s Cause Test Confirmed a Single Occurrence

Second, the court applied Ohio’s cause test directly, reiterating that Ohio determines the number of occurrences by reference to the cause or causes of the damage or injury, rather than by the number of individual claims, and that where a single, uninterrupted, and continuous cause produces the harm, all resulting injuries fall within that one occurrence.¹¹ ACE’s theory, the court observed, effectively treated the number of claims as the number of occurrences, precisely the equation Ohio’s cause test forecloses. ACE tried to escape that conclusion by relying on a line of Ohio asbestos-coverage cases, including LuK Clutch Systems, LLC v. Century Indemnity Co., in which courts examined whether a manufacturer’s business decision to use asbestos-containing components was the relevant occurrence for coverage-trigger purposes.¹² The Sixth Circuit found those cases inapposite: they addressed which policy period was triggered by a business decision made before the relevant coverage incepted, not whether the number of occurrences equals the number of retained-limit-triggering claims once coverage itself is undisputed. Because Smucker and ACE agreed that the salmonella contamination was accidental and occurred during the policy periods, the asbestos cases supplied no guidance.

The Lot Endorsement Was Ambiguous, Not a Multiplier

Third, and most consequential for the size of ACE’s exposure, the court addressed the Lot Endorsement, which stated that injury or damage arising from “the substantially same general harmful condition, cause, defect, error or suspected deficiency” and arising out of “any one ‘lot’” of Smucker’s product would be treated as a single occurrence. ACE read this language as converting what would otherwise be one occurrence into as many occurrences as there were implicated production lots, 225 of them. The Sixth Circuit disagreed, holding the provision ambiguous on two independent grounds. The endorsement never stated that it replaced the policies’ existing definition of “occurrence,” even though other endorsements in the same policies did say so expressly. And its central operative phrase, claims “arising out of any one ‘lot,’” was reasonably susceptible to either of two readings: as a limitation confirming that injuries within a single lot remain one occurrence, or, as ACE preferred, as a basis for aggregating injuries separately for each lot when multiple lots are implicated. Because Ohio law construes ambiguous policy language against the insurer that drafted it, the ambiguity favored Smucker.¹³ The court found particularly persuasive the Delaware Supreme Court’s decision in ConAgra Foods, Inc. v. Lexington Insurance Co., which addressed a nearly identical scenario, a salmonella-contaminated peanut butter recall and comparable lot-based aggregation language, and reached the same conclusion that the endorsement was ambiguous.¹⁴ The court also rejected ACE’s argument that Smucker’s reading rendered the endorsement superfluous, accepting Smucker’s alternative explanation that the endorsement performs a timing function, pinning an occurrence’s date to the first claim arising from a given lot so that claims traceable to one lot but spread across multiple policy years still trigger only a single retained limit.

Practical Implications for Insurers and Policyholders

First, policyholders facing a mass-contamination or mass-claim scenario should begin by identifying the operative cause of the alleged harm from their own perspective as the insured, not from the perspective of the individuals ultimately affected. Smucker confirms that under Ohio’s cause test, and under the closely related cause-based frameworks the Fifth and Ninth Circuits apply, the relevant inquiry is what the policyholder did or failed to do that unintentionally exposed it to liability, whether that is a contaminated production run, a defective component installed across many products, or a systemic process failure. Framing the claim around that unifying cause, rather than around the number of downstream injuries, is often the difference between a manageable retention and an existential one.

Second, insurers and policyholders alike should read aggregation endorsements, lot, batch, or similar provisions, with real skepticism about whose interest they actually serve. ACE assumed the Lot Endorsement worked in its own favor, multiplying Smucker’s retained limits on a per-lot basis and correspondingly shrinking ACE’s own exposure, when the same language could just as easily be read to consolidate claims that might otherwise be treated as separate occurrences. Smucker illustrates that lot and batch provisions are not inherently pro-insurer; their effect depends entirely on how “occurrence” is defined elsewhere in the policy and on whether the endorsement expressly overrides that definition.

Third, drafters on the insurer side should state explicitly whether an aggregation endorsement is intended to redefine “occurrence” or merely to clarify its application. The Sixth Circuit’s ambiguity finding rested in significant part on the fact that the Lot Endorsement never said it was replacing the policies’ existing definition, even though other provisions in the same policies did so. Insurers that want lot- or batch-based aggregation to multiply retained limits or erode aggregate coverage should draft language that unambiguously ties the number of occurrences to the number of lots, batches, or production runs, leaving no room for a competing reading.

Fourth, both sides should model the retention and limits math early, before litigation hardens competing occurrence theories. The gap between a single $250,000 retention and a $112.5 million aggregate retention is not a rounding error; it can determine whether coverage has any practical value at all in a mass-claim event. Policyholders facing a denial premised on a high occurrence count should press the insurer to justify that count against the policy’s own definition of occurrence, any aggregation language, and the cause test as applied in the governing jurisdiction, rather than accepting a claims-count shortcut.

Fifth, companies in food production, pharmaceuticals, consumer products, and other industries exposed to recall risk should revisit their CGL and umbrella towers with this decision in mind, particularly where those towers include lot, batch, or similar aggregation endorsements carried over from older policy forms. A provision that looks like boilerplate at renewal can become the single most consequential clause in the policy once a contamination event or systemic defect generates claims by the thousands.

A Cross-Circuit Perspective

Because Smucker arose under Ohio law, its holding does not bind courts outside the Sixth Circuit. But its reasoning tracks the cause-based approach that already prevails in the Fifth and Ninth Circuits, among others, suggesting that policyholders litigating mass-contamination or mass-recall coverage disputes in those jurisdictions should expect courts to ask the same threshold question Smucker asked: what did the insured do, viewed from its own perspective, that unintentionally caused the harm? At the same time, Smucker’s ambiguity holding on the Lot Endorsement is a reminder that outcomes in this area remain heavily dependent on specific policy language rather than on a jurisdiction’s general doctrinal label. Even in California, a cause-test state within the Ninth Circuit’s footprint, a state appellate court has remarked in a footnote that it believed a similar lot- and batch-aggregation clause was ambiguous as applied to the facts before it, an aside that did not drive the outcome there.¹⁵ Insurers underwriting recall or contamination risk in the Fifth and Ninth Circuits, no less than in the Sixth, cannot assume that a jurisdiction’s favorable general approach to “occurrence” will rescue a poorly drafted aggregation endorsement.

Conclusion

The Smucker decision does not change the underlying doctrinal test that Ohio, the Fifth Circuit, and the Ninth Circuit already apply to number-of-occurrences disputes. What it does is apply that test carefully to a fact pattern, a single contamination event radiating into thousands of claims, that is becoming more common as food safety recalls, product liability actions, and other mass-tort exposures grow in scale and public visibility. For policyholders, the decision is a significant win and a template for resisting claims-count-based occurrence theories. For insurers, it is a reminder that aggregation language must be drafted with precision if it is meant to multiply, rather than consolidate, the number of occurrences arising from a single mass-claim event.