Twenty years ago, in September 2006, Americans started getting sick from salad that came out of a bag. Not from a street cart or a questionable buffet. Washed, packaged baby spinach from the grocery store.
What makes this outbreak worth revisiting is not its size, though the size was considerable. It is that investigators followed the trail all the way from a bag in a victim’s refrigerator to a specific field, found the pathogen waiting there, and found no broken rule. That combination is what rewrote the industry.
The numbers everyone remembers, and the final ones
The figures that circulate online are 205 illnesses, 26 states, and three deaths. Those are real numbers: they are the snapshot from when the investigation closed, and they are what the 2007 publication reported.
The CDC’s final accounting came out in 2016, a decade later, and it reads differently. The total was 225 cases: 191 confirmed and 34 probable, across 27 states. Of those patients, 116 were hospitalized, which is 56 percent of everyone who got sick. Thirty-nine (19 percent) developed hemolytic uremic syndrome, a form of kidney failure. Five people died.
The gap between 205 and 225 is not anyone’s error. It is what every outbreak investigation looks like: the first counts are always incomplete, because probable cases have not been confirmed yet and some deaths come later. When you read a breaking news story about an outbreak, the number in the headline is the left-hand one.
From a bag to a field
The link to spinach came from interviews. Among 176 patients who carried the outbreak strain and could recall details, 161 (91 percent) had eaten fresh spinach in the ten days before they got sick. Among the 116 who named a brand, 106 (again 91 percent) had bought packaged product from one company, identified in the literature only as “brand A.”
Then came the step that turns statistics into evidence. Investigators collected bags out of the patients’ own refrigerators. They cultured the pathogen from 13 bags of brand A, and 12 of those isolates shared the same genetic fingerprint.
Traceback narrowed the product to a single production date, August 15, 2006, a single processing plant, and fields on four ranches on California’s central coast. On September 27, six weeks after that production date, the outbreak strain turned up in the environment for the first time. It was in cattle feces, roughly a mile from a spinach field on the property the reports call ranch A.
What they found at ranch A
A separate team went out and combed the property. They pulled 335 samples: feces from livestock and wildlife, surface water, sediment, soil, and well water. E. coli O157 grew from 45 of them, 13.4 percent. Twenty-eight of those isolates were indistinguishable from the outbreak strain.
Here is the picture that emerged. About 2,000 range cattle grazed the ranch. The spinach grew on a leased portion separated from the pastures by a wire mesh fence. Irrigation came from a well. And moving between the pasture and the field were feral swine.
The team counted them with traps and infrared cameras: roughly 149 animals on the ranch, a density of 4.6 swine per square kilometer. Forty were sampled. E. coli O157 turned up in 14.9 percent of the swine and 33.8 percent of the cattle. The swine reached the crop fields through gaps at the base of the fence, opened by their own rooting and widened by erosion.
Two traps sat 1.7 and 1.2 kilometers from the spinach field, and the animals visited them almost continuously from dusk to dawn, peaking between 5 p.m. and midnight. An average night brought 3.6 swine per trap. This was the first documented isolation of E. coli O157 from feral swine in the United States.
Two more numbers from that report are worth holding onto. The outbreak strain was still being recovered from the environment three months after the outbreak itself. And related isolates showed up in a river about 32 kilometers upstream, meaning the pathogen was living in the whole watershed rather than on one field.
What the investigation did not find
This is the uncomfortable part of the story, and the part most retellings skip.
- No manure-based fertilizer had been applied to that field.
- No runoff from the cattle pastures reached it.
- The topography ruled out flooding from surface waterways.
- The pathogen was never detected in the irrigation water at all.
- At the three other ranches in the traceback, the isolates did not match the outbreak strain.
There was no violated rule to point at. The fence was up. The water came from a well. No manure went on the field. And 225 people got sick anyway, five of them fatally. Where a rule was broken and the breach left a trail on paper, these stories end differently, with charges against named individuals. Three such US cases are collected in our piece on food safety prison sentences.
The authors are careful with their language: the mechanism of in-field contamination remains unclear, and a high density of feral swine near cattle and spinach fields could represent a risk factor. Not “was the cause,” but “could represent a risk factor.” That caution is appropriate, and it is worth preserving when the story gets retold.
Why one lot reached 26 states
A fecal deposit in a field is a local event. Turning it into an outbreak across twenty-seven states takes a conveyor belt.
Baby spinach is harvested with a machine much like a lawn mower. It cuts leaves in a continuous band and picks up whatever is lying in that band. The report says so directly: a harvester like that can pick up fecal deposits in the field and spread contamination across a large volume of product once it reaches processing.
After that, the math takes over. It takes remarkably few cells of E. coli O157 to make someone sick, which is why the infectious dose matters more than the total amount of dirt involved. Washing that bag would not have helped: some of the cells were already inside leaf tissue and under a film that no rinse removes.
The last link in the chain is pure logistics. A single production date is not one store; it is truckload after truckload moving across the country. That is why the outbreak map looked like the company’s distribution map instead of one county in California.
What grew out of it
In 2007, California growers created the California Leafy Green Products Handler Marketing Agreement, known as the LGMA. Its members now grow roughly 90 percent of the leafy greens in the United States, which works out to more than 50 billion servings a year. A nearly identical program operates in Arizona.
The structure is unusual, and the structure is the point. Joining the LGMA is voluntary. But a company that joins submits itself to mandatory government audits: USDA-certified inspectors, multiple times per season, with a requirement of 100 percent compliance rather than “mostly in order.”
On September 15, 2026, marking twenty years since the outbreak, the LGMA published a white paper on what has changed. The industry still dates its own history from that September.
What to take from this
First and most important: the contamination arrived from the field, not from anyone’s kitchen. The bag was already contaminated when it got to the store, and nothing a shopper did at home would have fixed it. “Wash your produce” is not the lesson here, and it is more honest to say so.
Second: safety in leafy greens rests on the things a checklist cannot catch. A fence that is standing is a box checked. A gap under that fence, opened by rooting and rain, is not a violation, it is just a gap. That is exactly why the LGMA built its program around audits in the field rather than around paperwork.
Third, the practical piece. If you grow greens yourself, the 2006 lesson applies to you directly: pay less attention to the tray and more to what walks around it and where the water comes from. How that breaks down by crop and by risk is collected in our guide to leafy greens food safety.
