The Dalles, 1984: How the Salad-Bar Outbreak Was Solved

AI-generated symbolic illustration of an empty restaurant. It does not depict a real venue or the 1984 events.

In September 1984, illness appeared among people who had eaten at restaurants in The Dalles, Oregon. By the end of the investigation, 751 people met the outbreak case definition, at least 45 had been hospitalised, and no death was reported. Cases rose in two waves. Ten of the town's 38 restaurants were definitely affected. Török and colleagues, JAMA, 1997, pp. 389–391.

The investigators did something precise: they linked illness to self-service salad bars. They did not initially identify who had contaminated them. That answer came after the collapse of the nearby Rajneeshpuram commune, when criminal investigators obtained witness accounts, purchase records and a laboratory sample matching the outbreak strain by the methods then used. Two senior commune figures later entered federal guilty pleas to conspiracy to tamper with consumer products. Török et al., pp. 390, 393–394; United States v Sheela, 667 F Supp 724, 725 (1987).

The essential distinction is between finding the exposure, attributing the act, and establishing an individual's legal responsibility. The first happened in 1984. The decisive criminal evidence followed in 1985. The pleas came in July 1986. The label “Rajneeshee attack” is common, but it must not imply that every member knew of or took part in the crime.

Followers of the Indian spiritual teacher Bhagwan Shree Rajneesh bought a ranch in Wasco County in 1981 and built Rajneeshpuram. The community's growth and attempts to incorporate a city brought land-use litigation and conflict with local authorities. Its leadership also sought influence over the November 1984 county election. None of those disputes made illness an inevitable tactic; they explain the political setting that later witnesses described. Carl Abbott, Oregon Historical Society, “Rajneeshees”; Török et al., p. 390.

The criminal inquiry's informants said the restaurant contamination was a test of whether enough local voters could be made ill to alter participation in the upcoming election. The investigators recorded plans concerning the city's water supply, but their outbreak study found no evidence of a waterborne outbreak. A reported plan is not an executed town-wide attack. Carus's later case study reports that the larger November plan was abandoned; the published outbreak study documents no repeat of the September restaurant event. Török et al., pp. 393–394; Carus, printed p. 58.

The Wasco-Sherman health department began receiving restaurant-linked reports on 17 September 1984. Local and state officials confirmed Salmonella Typhimurium. As more cases appeared, the department closed the town's salad bars on 25 September and Oregon sought CDC assistance. The published retrospective counted onset dates from 9 September to 10 October, with a first wave through 18 September and a larger second wave from 19 September. Török et al., pp. 389–391.

Investigators compared what sick and well customers had eaten and interviewed employees. Eating from salad bars was the consistent risk in the four restaurants where detailed customer comparisons were possible. Eight of ten definitely affected restaurants operated salad bars, compared with three of 28 other restaurants. Different food items were implicated at different venues; no one food supplier or distributor linked the affected restaurants. Water systems did not fit the pattern. These were observations about exposure, not a direct identification of the contaminator. Two of the ten affected restaurants did not have salad bars, and the paper allows that illness could spread secondarily in some settings. Török et al., pp. 389–393.

The case count needs care. Of the 751, 388 were culture-confirmed and 363 met clinical criteria. Eleven were classified as secondary cases, while 48 had incomplete restaurant-exposure information. The number is a defensible outbreak total under the authors' case definition, not a claim that all 751 had laboratory confirmation or that every ill person was found. Török et al., table, p. 391.

Food handling problems were real. Some employees became ill and may have passed infection to food. Inspectors found minor hygiene lapses, and the way some salad bars were maintained may have helped illness spread. The investigators initially considered intentional contamination but found no known motive, no claim of responsibility and no clear pattern of suspicious customers. They expected one act of sabotage, while the exposure curves suggested repeated events. They therefore favoured more familiar accidental mechanisms at the time. Török et al., pp. 393–395.

Those mechanisms did not fully explain the starting point. There was no common supplier across the implicated restaurants, private banquets in affected venues were spared, and the implicated foods varied. The published paper says infected staff may have contributed to later transmission but found no evidence that they initiated the outbreak. That is a lesson in inference, not a claim that public-health investigators ignored an obvious culprit: a pattern can identify where exposure happened while still leaving agency and motive beyond epidemiology's reach. Török et al., pp. 389, 393–395.

After senior Rajneeshpuram figures left in September 1985, Rajneesh accused his former secretary Ma Anand Sheela and others of crimes and invited investigation. His claims alone were allegations. Investigators searched the commune's medical facilities on 2 October 1985. Clinic records showed that a Salmonella Typhimurium strain had been obtained before the outbreak; the sample found there was indistinguishable from the outbreak strain on the biochemical, antibiotic-sensitivity and plasmid tests reported in the paper. Informants supplied accounts of deliberate contamination and the proposed election motive. Török et al., pp. 390, 393–394.

This is converging evidence, not a miracle fingerprint. The researchers found similar profiles in several other Oregon outbreaks and one animal isolate without an established link to The Dalles. Informant recollections could not be matched neatly to every restaurant's exposure dates. Those limits matter. The attribution rests on the combined epidemiology, the commune sample and purchase timing, the witness accounts, and the subsequent pleas. A matching strain by itself would not identify a person. Török et al., pp. 393–394.

The federal court opinion in United States v Sheela states that Ma Anand Sheela and Diane Onang, known as Ma Anand Puja, entered guilty pleas on 22 July 1986 to conspiracy to tamper with consumer products and illegal wiretapping. Both received four-and-a-half-year federal sentences under negotiated agreements. The opinion concerned a later dispute about post-sentence reports; it confirms the pleas and sentences but did not try every factual allegation in the story. United States v Sheela, 667 F Supp 724, 725 (1987).

Contemporary reporting records separate state pleas and sentences concerning an attempted murder, assaults and arson. These were distinct offences, and a long nominal state sentence should not be substituted for the federal term in the food-tampering case. The full plea transcripts and original indictment were not available in the routes checked for this investigation. A later television interview reported Sheela denying that she caused the outbreak and saying she pleaded guilty because she expected no fair trial. That is her attributed position; it does not undo the recorded guilty plea, the laboratory findings or the published epidemiological assessment. Associated Press, 23 July 1986; UPI, 3 August 1987.

Rajneesh's own federal case ended in a separate immigration plea in November 1985. That outcome cannot be used as proof that he ordered or knew of the restaurant attack. There is contrary evidence to weigh: historian W. Seth Carus reports a 1985 interview in which former commune mayor David Berry Knapp said Sheela had told him she discussed the voter-illness plan with Rajneesh. Knapp was reporting Sheela's alleged account of a private conversation, not a conversation he witnessed; the original interview and independent corroboration were not available in this review. OSHO World's sympathetic biographical account attributes crimes to Sheela's management and says Rajneesh invited investigators. Neither Knapp's second-hand allegation nor OSHO World's account settles his personal knowledge. Sheela and Onang's food-tampering pleas are established; Rajneesh's role in the attack remains unresolved on the record examined here. Carus, Bioterrorism and Biocrimes, printed p. 52 and note 162; OSHO World, “About Osho”; Abbott.

The lesson is about the boundary between disciplines. Epidemiology measured a real outbreak and identified the likely exposure route. Criminal investigation established purposeful action and supplied a motive that the health data could not. In the authors' own account, the possibility of intention was considered early, then set aside for reasons that seemed sensible until later evidence arrived. Their recommendation was to maintain ordinary outbreak surveillance and seek an independent law-enforcement investigation when a large, puzzling pattern resists established explanations. They did not argue that every unusual cluster is an attack. Török et al., pp. 394–395.

The harms extended beyond the 751 counted cases: hospital and laboratory work rose sharply, and restaurants and residents lived with fear and disruption. One later specialist review says the wider economic cost cannot be measured reliably from the available data. Claims of a precise town-wide financial toll would therefore overstate the record. Chapman and Zilinskas, Nuclear Threat Initiative, 2007.

Verified: The Dalles experienced a two-wave restaurant-linked Salmonella Typhimurium outbreak in 1984; 751 people met the published case definition, at least 45 were hospitalised and no deaths were reported. Self-service salad bars were the principal measured risk. The 1985 criminal inquiry found a matching strain in the commune clinic and gathered accounts of deliberate contamination. Sheela and Onang entered 1986 federal guilty pleas to consumer-product tampering conspiracy.

Contested or limited: Sheela later denied causing the outbreak. The initial food-handler explanation could account for some later spread, but the full evidence does not support it as the origin of the multi-restaurant outbreak. The strain comparison corroborates the criminal case; it does not uniquely identify every act or actor.

Unresolved: The precise role of every participant, the relationship of each alleged peripheral attempt to the measured outbreak, and Rajneesh's personal knowledge or approval. Knapp's second-hand account raises a serious question but cannot settle it. Original plea transcripts, the complete Knapp interview and any direct contemporaneous record of Rajneesh's knowledge could refine those answers. An immigration plea cannot fill that gap.

Method note

This article reviewed the full 1997 epidemiological paper, the federal court's 1987 opinion, contemporary wire-service accounts, Carus's historical synthesis, the Oregon Historical Society's overview, and OSHO World's sympathetic account. The article and duplicate-publication searches were refreshed on 25 October 2026. The original indictment, plea transcripts and Knapp interview were not recovered; the government scan of the Attorney General's 1986 report could not be independently checked. No operational contamination detail is reproduced here. Sources are classified by what they directly establish; accounts that repeat the same inquiry are not treated as independent confirmations.