Every summer somebody asks me what the biggest foodborne illness outbreak in American history was, and every summer I have to answer the question with a question: biggest how?
Because “largest” is doing at least four different jobs. There is the outbreak that made the most people sick. There is the outbreak that killed the most people. There is the outbreak that pulled the most product off the most shelves. And there is the outbreak that actually changed something. Those are four different lists, and the same outbreak almost never tops more than one of them. Thirty-three years of this work has taught me that the gap between those lists is the whole story.
So here is the catalogue, as complete as I can make it, with the numbers as the agencies and the journals finally settled them rather than as the headlines first reported them. Read it with the Cyclospora outbreak we are living through right now held in your other hand.
Largest by the number of people made sick
Schwan’s ice cream, 1994. Salmonella Enteritidis. Marshall, Minnesota. This is still the largest common-source Salmonella outbreak ever recognized in the United States, and almost nobody remembers it. The New England Journal of Medicine investigation estimated that 224,000 people nationwide developed salmonellosis after eating Schwan’s ice cream, with a consumer attack rate of 6.6 percent. The vehicle was not the ice cream itself but the premix: tanker trailers hauled the pasteurized base to the Marshall plant after hauling unpasteurized liquid egg, and nobody re-pasteurized after transport. Investigators isolated the outbreak strain from 8 of 266 ice cream products but from none of the 157 environmental samples taken inside the plant and none of the 204 taken from the tankers. In 1994 the CDC logged 10,009 Salmonella Enteritidis reports nationally, a 21 percent jump over the year before, and this single product was a meaningful share of it.
I was asked to take those cases. I turned them down. In the fall of 1994, I was up to my eyeballs in Jack in the Box, and I did not have another 224,000-person outbreak in me. It is the only time I have ever passed on the largest outbreak in the country, and I have thought about it more than once since.
Hillfarm Dairy and Jewel Food Stores, 1985. Salmonella Typhimurium, antimicrobial-resistant. Melrose Park, Illinois. Two separate waves of illness, more than 16,000 culture-confirmed cases, traced to two brands of pasteurized two-percent milk from a single dairy plant. Two independent surveys estimated the true number of people affected at 168,791 and 197,581, which by the estimated count makes it the largest outbreak of salmonellosis ever identified in this country. Illness was associated with having taken antimicrobials before onset — the resistant strain had a clear field. A pasteurized product, from a licensed plant, in a major grocery chain. That was the point of the story then and it is the point now.
Sakai City, Japan, 1996. E. coli O157:H7. White radish sprouts from a single farm, served in school lunches. The World Health Organization reported that as of August 26, 1996 the Sakai outbreak had affected 6,309 schoolchildren and 92 school staff across 62 municipal elementary schools, with roughly 160 additional cases mostly among family members. Hospitalizations peaked at 534 on July 18. Hemolytic uremic syndrome peaked at 101 cases on July 24. Two girls, ages 10 and 12, were dead by the end of August. Nationwide, Japan logged 9,578 O157:H7 cases and 11 deaths that summer; the CDC counted roughly 10,000 cases across at least 14 separate clusters between May and August, most of them in school-age children. Japanese accounts have since revised the Sakai figure upward to 9,523 sick, including 7,892 elementary schoolchildren, with 12 deaths overall. The epidemiology is worth reading in full: sprouts were the only uncooked food common to the implicated meals in two school districts, and the same farm’s sprouts were tied to a factory cafeteria outbreak in Kyoto 50 kilometers away. One of the Sakai children died of the sequelae of her infection twenty years later.
China, 2008. Melamine, not a pathogen — which is exactly why it belongs on this list. Melamine was deliberately added to diluted raw milk at collection stations to inflate the apparent protein content, and it ended up in infant formula. The Chinese Ministry of Health confirmed 294,000 cases and six deaths as of December 1, 2008, out of 22.4 million children screened, with roughly 51,900 hospitalized for kidney stones and renal injury. Twenty-two dairy companies were implicated, with the Sanlu Group at the center. It remains the largest deliberate food contamination event ever documented, and the largest food safety event of any kind in the last fifty years by number of people harmed.
Shanghai, 1988. Hepatitis A virus in raw clams. This is the largest food-associated outbreak of any kind ever recorded, and it barely appears in American food safety writing. Clams harvested from a newly opened, sewage-polluted bed were shipped into a city where most young adults had no immunity, and were eaten raw or barely cooked. The published epidemiology counts 292,301 cases at an attack rate of 4,083 per 100,000 over roughly two months in the first quarter of 1988. Among people who ate clams the attack rate was 11.93 percent; among those who did not, 0.52 percent. Forty-seven people died, with elevated mortality among patients who were already hepatitis B positive. It remains the largest virus-associated foodborne outbreak ever reported. One contaminated harvest area, one uncooked shellfish habit, and a quarter of a million people with hepatitis in eight weeks.
Deadliest
South Africa, 2017–18. Listeria monocytogenes sequence type 6 in polony, a ready-to-eat processed meat, made at the Enterprise Foods plant in Polokwane, owned by Tiger Brands. Between January 1, 2017 and July 17, 2018 South Africa recorded 1,060 laboratory-confirmed cases and 216 deaths. Nothing else in the history of foodborne illness comes close: it is the largest and deadliest listeriosis outbreak ever documented anywhere in the world, in a country that had previously logged 60 to 80 cases a year. Neonates 28 days old or younger accounted for 43 percent of the cases, and among the 806 patients whose outcome was known, 27 percent died. Whole genome sequencing did the work: 91 percent of clinical isolates were ST6, matching isolates from the polony and from swabs of the factory environment, and inspections of all 157 ready-to-eat meat facilities in the country found ST6 in exactly one of them. When the health minister named the plant, the Tiger Brands share price fell seven percent, and the chief executive told the room there was no direct link between any death and his company’s products, then declined to apologize. Our firm has assisted the South African class action. Nine years on, the claimants have not been compensated. Two hundred sixteen people.
Jalisco Mexican Products, 1985. Listeria monocytogenes in Mexican-style fresh soft cheese, Artesia, California. Still the deadliest listeriosis outbreak in American history. In Los Angeles County alone, 142 cases were reported between January 1 and August 15, 1985. Ninety-three of them — 65.5 percent — were in pregnant women or their infants. There were 48 deaths in the county series: 20 fetuses, 10 newborns, and 18 nonpregnant adults, with the commonly cited national total at 52. Eighty-seven percent of the maternal and neonatal cases were in Hispanic families. This was the outbreak that taught American public health what Listeria does to pregnancy, and it is the reason every obstetrician in the country now hands out a list of cheeses to avoid.
Bil Mar Foods and Sara Lee, 1998–99. Listeria monocytogenes in hot dogs and deli meats out of Zeeland, Michigan. One hundred one illnesses, 21 deaths, and six miscarriages or stillbirths. The CDC isolated the outbreak strain from both an opened and a previously unopened package of hot dogs made at the plant. The recall began at 15 million pounds in December 1998 and grew to 35 million — Ball Park, Bil Mar, Bryan, Grillmaster, Hygrade, Mr. Turkey, Sara Lee. The likely mechanism was mundane and awful: a refrigeration unit was removed from the building in sections, and moving the pieces appears to have spread Listeria through the plant. A federal meat inspector later told investigators that management knew of elevated Listeria levels about eight months before the outbreak and shipped anyway. Bil Mar pleaded to a single misdemeanor count in 2001: a $200,000 fine and $3 million for food safety research at Michigan State. Twenty-one people, $200,000.
Maple Leaf Foods, 2008. Listeria monocytogenes in cold cuts from the Bartor Road plant in North York, Toronto — establishment 97B. Fifty-seven confirmed cases and 23 deaths, though the figure is reported variously as 22, 23, or 24 depending on whether you are reading the Weatherill inquiry, the Public Health Agency, or the peer-reviewed paper, which put it at 24 deaths among 57 cases — a 42 percent case fatality rate. Seventy-two percent of the people infected were residents of long-term care facilities or hospital inpatients at the time of exposure, which is the whole tragedy of Listeria in one statistic: it finds the people least able to survive it. The organism was traced to organic material lodged deep inside two meat slicers on lines 8 and 9, surviving a sanitation regime that met the equipment manufacturer’s own specifications. Maple Leaf recalled every product made at the facility — more than 200 items — and closed the plant. Michael McCain went in front of the cameras and said his company’s products had been linked to illness and loss of life, offered his sympathies to the families, and did not route it through counsel. It is still taught as the model of how to answer for an outbreak, and it stands in useful contrast to the Tiger Brands press conference above. It did not save the plant from CFIA inspectors finding mold, slime and rusty equipment there a month after it reopened.
Jensen Farms, 2011. Listeria monocytogenes in whole cantaloupe marketed as Rocky Ford, grown at Holly, Colorado and packed at Granada. The CDC’s final tally was 147 cases in 28 states, 143 hospitalizations, 33 deaths and one miscarriage, with Colorado carrying 40 of the illnesses and nine of the deaths. It was polyclonal — four, later five, distinct PFGE patterns across serotypes 1/2a and 1/2b, which is not how a single-point contamination behaves and tells you the organism had multiple niches in that building. The FDA found used equipment bought in July 2011 that had previously run a different commodity, corroded and hard to clean, floors that pooled water, and no antimicrobial wash on the melons. Roughly 300,000 cantaloupes shipped between July 29 and September 10 to at least 24 states. In 2013 Eric and Ryan Jensen were charged with six counts of introducing adulterated food into interstate commerce — strict-liability misdemeanors under the Food, Drug, and Cosmetic Act, no proof of intent required. I represented families in this one. Thirty-three people died and the brothers got probation and home detention.
Germany, 2011. E. coli O104:H4 in fenugreek sprouts grown from Egyptian seed. The Robert Koch Institute’s final accounting was 855 HUS cases and 2,987 cases of EHEC gastroenteritis without HUS — 3,842 in Germany, with the broader European figure usually given as 3,950 — and 53 deaths, 35 of them among the HUS patients. It remains the largest HUS outbreak ever described anywhere in the world. What made it strange was the demographics: median age 46 among the gastroenteritis cases and 42 among the HUS cases, with 68 percent of HUS patients female. HUS is supposed to be a disease of small children. This was an adult epidemic, and the strain was a hybrid — Shiga toxin from enterohemorrhagic E. coli bolted onto the adherence machinery of enteroaggregative E. coli. Investigators worked backward through 41 separate outbreak clusters to a single horticultural operation in Lower Saxony. They never isolated the organism from the seed.
Boar’s Head, 2024. Listeria monocytogenes in liverwurst and deli meats from Jarratt, Virginia. Sixty-one people sick in 19 states, 60 of them hospitalized, ten dead — the deadliest listeriosis outbreak since Jensen Farms. The company recalled more than seven million pounds across 71 products after contamination was found on a pallet jack moving between the raw and ready-to-eat sides of the plant. USDA inspectors had logged 69 regulatory violations at that facility between August 2023 and August 2024, documenting black mold, mildew, insects, and blood pooling on the floor. The plant is closed. Liverwurst production ended permanently. I am still working these cases.
Largest by recall and market disruption
Peanut Corporation of America, 2008–09. Salmonella Typhimurium in peanut butter and peanut paste from Blakely, Georgia. Seven hundred fourteen cases in 46 states, 166 hospitalizations, nine deaths. The illness count is not what makes this the recall of the era. PCA sold an ingredient, so the contamination propagated into 3,918 recalled products from more than 200 manufacturers — crackers, cookies, ice cream, pet treats, school snacks — and the FDA had to stand up a searchable website because the recall list was too long to communicate any other way. PCA’s customers put their losses at roughly $144 million. The company filed Chapter 7 within weeks. Then something happened that had never happened before: the government tried a food poisoning case to verdict. Stewart Parnell was convicted on 71 counts including conspiracy, obstruction, and wire fraud, and sentenced to 28 years; his brother Michael got 20. We filed nine lawsuits and represented over 100 clients, and two of our families testified before the House Oversight and Investigations subcommittee in February 2009.
Wright County Egg and Hillandale Farms, 2010. Salmonella Enteritidis in shell eggs from Iowa. 1,939 confirmed cases and no confirmed deaths, but the FDA estimated the real number in the tens of thousands, and the recall reached about 550 million eggs — 380 million from Wright County Egg, 170 million from Hillandale. Congressional investigators later found that Wright County Egg had received 426 positive Salmonella results between 2008 and 2010, including 73 potentially positive for Enteritidis, one of them on July 26, less than three weeks before the recall. Jack DeCoster had been designated a habitual violator of Iowa environmental law in 2000. When the two men were called before the House Energy and Commerce Committee, one refused to answer and the other took the Fifth. We filed the first lawsuit in the outbreak on August 17, 2010.
Dole and Natural Selection Foods, 2006. E. coli O157:H7 in prepackaged baby spinach out of San Juan Bautista, California. The CDC’s count was 199 confirmed cases in 26 states, 102 hospitalized, 31 with HUS, three dead; the final FDA and California report put the illness figure at 205. Two elderly adults and a two-year-old with HUS died. The traceback landed on the Paicines Ranch in San Benito County, where investigators found the outbreak strain in river water, in cattle feces, and in feral pig feces, with a grass-fed cattle operation less than a mile from the spinach field. The market effect was total: there were no fresh spinach sales anywhere in the United States for five days, and California spinach stayed off the market for ten days after that. I represented 93 people in that outbreak. It produced the California Leafy Greens Marketing Agreement, which was the industry’s answer to being regulated, and which has now failed to prevent leafy green outbreaks for nineteen consecutive years.
Yuma romaine, 2018. E. coli O157:H7. Two hundred ten people in 36 states, 96 hospitalized, 27 with HUS, five dead in Arkansas, California, Minnesota (two) and New York. Nearly half of everyone identified went to a hospital. It was the largest O157:H7 outbreak in the country since the 2006 spinach outbreak, and the CDC found the outbreak strain in canal water in the Yuma growing region — a 3.5-mile stretch of the Wellton irrigation canal serving 23 farms, with a concentrated animal feeding operation nearby. The FDA’s environmental assessment said the quiet part out loud: because romaine from many farms was commingled at the processors, it was impossible to determine which farm supplied the contaminated lettuce.
Salmonella Saintpaul, 2008. This one earns its place for a reason that has nothing to do with pounds recalled. Between April and August 2008, 1,442 cases were confirmed across 43 states, the District of Columbia and Canada, with 21 percent hospitalized and two deaths. Early case-control work pointed hard at raw tomatoes. The FDA warned the public off tomatoes, tomato consumption collapsed, and the tomato industry lost an estimated $200 million — for a product that was never microbiologically confirmed as the source and whose traceback never converged on anything. The outbreak strain was eventually isolated from jalapeño peppers in Texas and from serrano peppers and agricultural water on a single Mexican farm. The peppers were probably eaten with the tomatoes, which is why the early studies pointed the wrong way. Read that again with this summer in mind: a category-wide warning, a wrecked commodity, a traceback that could not resolve to a farm, and an imported product from Mexico as the eventual answer.
A few more belong on the ledger even if they do not top a column. Pilgrim’s Pride, 2002 — Listeria in sliced turkey, 54 sick, 8 dead including three stillbirths, and 27.4 million pounds recalled, the largest meat recall in the country at the time. Cargill Turkey Products, 2000 — 29 sick, 7 dead, three of them miscarriages or stillbirths. Odwalla, 1996 — unpasteurized apple juice, 66 sick, a 16-month-old girl dead, and the outbreak that produced juice HACCP. Blue Bell, 2015 — Listeria in ice cream, 10 sick and 3 dead from a product nobody had thought of as a Listeria vehicle, and a rare criminal prosecution of a chief executive. Westland/Hallmark, 2008 — 143 million pounds of beef, the largest meat recall in American history, with zero confirmed illnesses, which tells you that recall size and human harm are almost unrelated quantities. And Milwaukee, 1993 — 403,000 people with cryptosporidiosis from the municipal water supply, which is not foodborne but is the only American event of the last fifty years that operated at the scale of the Shanghai and Chinese formula numbers above.
Largest by consequence
Jack in the Box, 1992–93. E. coli O157:H7 in undercooked hamburger. Seven hundred thirty-two people sick, 178 with permanent injury, four children dead. By the raw numbers it does not crack the top of any list above. It is not in the same universe as Schwan’s. It killed a fifth as many people as Jalisco. And yet it is the most consequential foodborne outbreak in American history by a distance nothing else approaches. It made E. coli O157:H7 an adulterant in ground beef in 1994. It produced pathogen testing, HACCP in meat plants, the modern cooking temperature standard, and PulseNet. It built the entire architecture that every other outbreak on this page was detected by. It also built my law firm, and it is the reason I have spent thirty-three years doing this instead of something easier.
So, look at the arithmetic of that. Seven hundred thirty-two people changed federal law. Two hundred twenty-four thousand people did not change anything anyone can name. Five hundred fifty million eggs and 426 positive lab results produced no federal charge for the outbreak itself. Twenty-one deaths at Bil Mar cost Sara Lee little. Thirty-three deaths at Jensen Farms cost the Jensen brothers their probation officers’ time. The relationship between how many people an outbreak hurts and what an outbreak changes is not weak. It is close to nonexistent. What determines consequence is whether somebody counts the sick, names the company, and makes it cost something.
Which brings me to the outbreak on my desk this morning. As of its July 14 health advisory, the CDC has 1,645 laboratory-confirmed domestically acquired cases of cyclosporiasis across 34 states since May 1, plus more than 5,100 additional cases it says require further analysis, against 249 cases in the same window last year. Of the confirmed patients, 141 — nine percent — have been hospitalized. The Taco Bell cluster the FDA and CDC named on July 16 accounts for at least 1,644 of those people across five states, tied to shredded iceberg lettuce from Taylor Farms de Mexico. Michigan alone reported 4,312 outbreak-period cases and 102 hospitalizations as of July 16, and its year-to-date total has since passed 5,000. Ohio has reported more than 1,200 cyclosporiasis cases this year.
Run those numbers against the list above. If you count the way Michigan counts, this outbreak is already among the largest foodborne outbreaks in modern American history — bigger than Yuma romaine, bigger than Dole spinach, bigger than PCA, bigger than Wright County Egg, bigger than every Listeria outbreak on this page put together, several times over — including the largest one the world has ever recorded. If you count the way the CDC counts, it is 1,645 people. Same outbreak. Same summer. The difference is not epidemiology. It is bookkeeping.
And the bookkeeping was a choice. On July 1, 2025, FoodNet made Cyclospora surveillance optional, leaving mandatory tracking for exactly two pathogens. Then we walked into the biggest cyclosporiasis season on record with the instrument turned off. The CDC itself says the true number is likely higher than what has been reported that the outbreak may not be limited to the states with known illnesses, and that it can take six weeks to determine whether a sick person belongs to it. Meanwhile Taylor Farms’ own recall notice lists shredded iceberg distributed to 27 states between June 29 and July 16, including retail consumer sizes, while the federal advisory covers five states and says grocery product is not implicated.
Every outbreak on this list was invisible until somebody counted it. The 224,000 people who ate Schwan’s ice cream were invisible until a Minnesota epidemiologist noticed her numbers were wrong. The 168,000 people who drank Jewel milk were invisible until somebody ran two surveys. Yuma romaine went unattributed to any single farm because the product was commingled and nobody had lot-level records. That is the same failure we are watching in real time, with the same commodity, from the same region, through the same distributor, and there is a rule already written to fix it — FSMA 204, finalized in November 2022, with leafy greens on the Food Traceability List and a 24-hour requirement to produce lot-level records. Compliance was set for January 2026. The FDA proposed pushing it to July 2028 and Congress directed non-enforcement in the meantime.
Fifty years of outbreaks and the pattern does not move. We find out how big it was afterward, from a journal article, years later, when it is too late for the rule to matter and too late for the family. The one variable we control is whether we are willing to count while it is happening. Right now, we are not.