Cookie dough, flour and the snack you ate instead

The largest known outbreak traced to raw cookie dough made 77 people ill across 30 states, put 35 in hospital and left 10 with haemolytic uraemic syndrome. Of the case patients, 94 percent had eaten ready-to-bake prepackaged dough against 11 percent of controls, a matched odds ratio of 41.3, and no other exposure came close. The eggs were never the problem. Investigators found three further toxin-producing E. coli strains in the implicated brand, and 3.6 million packages were recalled. Years later a national survey of 2 171 American adults found 35 percent had tasted something with uncooked flour in the past twelve months, and on average they rated uncooked flour unlikely to carry germs that make people sick. The baked cookie raises a different question, and a smaller one. Baking creates acrylamide in wheat foods, pooled at 136 µg/kg across 28 studies, and Europe set 350 µg/kg for biscuits and wafers as a level bakers are expected to stay under rather than a legal limit.

The numbers, per 100 g

How much is actually in it

• Researchers picked sweet biscuits as the yardstick snack because national survey data said that is what people actually eat. — New Zealand adult population data used to select five commercial biscuit varieties (n = 5 biscuit varieties) Evidence B source
  The control snack was made up of named commercial biscuits and sized to the greater of 10% of total energy requirements or the energy in 42.5 g of almonds, which is 1030 kJ

What a real portion looks like

• Swapped calorie for calorie against almonds, sweet biscuits raised blood glucose more and left people eating more later. — 100 adults with available data, 25 men and 75 women, out of 106 randomised (n = 100) Evidence B source
  Glucose area under the curve was 53 mmol/L.min lower for almonds (95% CI 45 to 61, p < 0.001); energy intake over the rest of the day was 638 kJ lower after almonds (95% CI 44 to 1233, p = 0.035); appetite ratings differed only at 90 minutes and lunch intake did not differ

What cooking and storage change

• Baking itself creates the contaminants measured in cookies, and acrylamide is the one everyone measures. — 28 studies of wheat-based foods including bread, cookies and pasta (n = 28 studies) Evidence C source
  Acrylamide was assessed in 69.7% of studies and reached the highest pooled concentration at 136.29 µg/kg, followed by hydroxymethylfurfural at 70.59 µg/kg and polycyclic aromatic hydrocarbons at 0.11 µg/kg; only four studies on cookies measured HMF, and no commercial-sample data exist for some regions of the world
• Europe never set a legal limit for acrylamide in biscuits, it set a number bakers are expected to get under. — EU food business operators producing fine bakery wares Evidence E source
  The benchmark level for biscuits and wafers is 350 µg/kg, alongside crackers other than potato-based at 400, crispbread at 350 and gingerbread at 800
• Eating fruit half an hour before a wheat biscuit nearly halved the height of the glucose rise. — 20 subjects (n = 20) Evidence B source
  Every exchange cut the glycaemic incremental area under the curve by 20 to 30% with no loss of total satiation; the peak amplitude fell almost 50% when the fruit came 30 minutes ahead of the biscuit, and less when it came with it or after

What it does to your health

• Biscuits built as a delivery vehicle cut anaemia in schoolchildren against the control group, while the weekly supplement group fell in between. — 403 Vietnamese school children (n = 403) Evidence B source
  Anaemia prevalence after six months was 1.0% in the fortified biscuit group against 10.4% in control (P = 0.006), with the weekly supplement group intermediate at 7.4%; both intervention groups improved iron status, and vitamin A deficiency at baseline modified the effect so that haemoglobin rose in deficient children given biscuits but not in those given the supplement

What people believe that the data does not support

• The biggest known outbreak from raw cookie dough came from a factory-sealed pack, and the eggs were never the problem. — 77 patients from 30 states with illness onset between 16 March and 8 July 2009, 35 of them case patients in the matched analysis against 36 controls (n = 77 cases) Evidence C source
  33 of 35 case patients (94%) had eaten ready-to-bake prepackaged cookie dough against 4 of 36 controls (11%), matched odds ratio 41.3, P < .001, with no other exposure significant; 35 were hospitalised, 10 developed haemolytic uraemic syndrome, 66% were under 19 and 71% female; three non-outbreak STEC strains were isolated from the implicated brand and 3.6 million packages were recalled
• A third of American adults had eaten something with raw flour in the past year and rated it unlikely to make anyone sick. — 2171 US adults (n = 2171) Evidence D source
  35% reported tasting or eating something containing uncooked flour in the previous 12 months, with significant differences by sex, race, education and age; on average respondents said uncooked flour is not likely to contain germs that can make people sick, and rated raw homemade cookie dough only moderately likely to

Who this works differently for

• Higher acrylamide intake in pregnancy tracked with a small rise in the odds of a baby born small for dates. — 1471 mother-child pairs in the EDEN cohort (n = 1471) Evidence C source
  Median intake 19.2 µg/day (IQR 11.8 to 30.3); each 10 µg/day higher intake carried odds ratio 1.11 (95% CI 1.03 to 1.21) for small for gestational age, birth length -0.05 cm (95% CI -0.11 to 0.00) and birth weight -9.8 g (95% CI -21.3 to 1.7)

The bottom line

Do not eat raw dough, and know that the flour is the reason rather than the egg. The bowl and the beaters count too. That instruction costs you nothing, and the outbreak behind it hospitalised 35 people, two thirds of them under 19 and most of them female. On the baked cookie the numbers are smaller than the worry. Acrylamide in wheat foods pools at 136 µg/kg against a European benchmark of 350 for biscuits, 400 for crackers and 800 for gingerbread. In 1 471 French mother and baby pairs, each extra 10 µg of dietary acrylamide a day in pregnancy carried an odds ratio of 1.11 for a baby born small for dates. The median intake in that cohort was 19.2 µg a day, and the birth weight difference was 9.8 g with a confidence interval that crossed zero. That study is a cohort, which is why it carries a C. It records an association rather than an effect, and nobody measured acrylamide in anyone. It was estimated by multiplying a food questionnaire by national contamination averages, so two layers of error sit under the number. Read it as a reason for a pregnant woman to keep browned starchy food at ordinary levels, and not as a reason for anyone to count biscuits. The comparison that changes a day is what you eat instead. Matched calorie for calorie against almonds, sweet biscuits raised the blood glucose area under the curve by 53 mmol/L.min and left people eating 638 kJ more over the rest of the day, with appetite ratings that barely differed. The researchers chose biscuits as the yardstick because national survey data said that is what people snack on. One last thing to keep in proportion. A biscuit built as a delivery vehicle, fortified with micronutrients and fed daily to Vietnamese schoolchildren for six months, cut anaemia from 10.4 percent in the control group to 1.0 percent. The children given a weekly iron pill instead landed in between at 7.4 percent, and the trial compared each group with the control rather than the biscuit with the pill, so it does not say the biscuit beat the supplement. What it tells you is what a biscuit can carry, and nothing about the one in your cupboard.

Sources

1. coo-r5-07 · test food specification within a randomised crossover trial
   Sweet biscuits were chosen as the control as these were shown to be one of the most commonly consumed snack foods in the latest adult national nutrition survey in New Zealand
   Int J Environ Res Public Health, 2021 · checked 2026-09-22
2. coo-r5-06 · acute randomised controlled crossover trial
   The mean area under the blood glucose response curve was statistically and practically significantly lower for almonds than biscuits (mean (95% CI) difference: 53 mmol/L.min (45, 61), p < 0.001).
   Int J Environ Res Public Health, 2021 · checked 2026-09-22
3. coo-r5-03 · systematic review and meta-analysis of occurrence data
   The highest concentration was for acrylamide (136.29 µg·kg-1) followed by HMF (70.59 µg·kg-1) and PAHs (0.11 µg·kg-1).
   Crit Rev Food Sci Nutr, 2023 · checked 2026-09-22
4. coo-r5-05 · EU regulation, benchmark levels annex
   Biscuits and wafers 350
   Commission Regulation (EU) 2017/2158, Annex IV, benchmark levels for acrylamide · checked 2026-09-22
5. coo-r5-09 · randomised crossover intervention study
   Glycaemic response amplitudes were reduced almost 50% compared with 74 g WB when KF ingestion preceded WB ingestion by 30 min, and less when the KF was ingested with or 30 min after the cereal.
   Nutrients, 2023 · checked 2026-09-22
6. coo-r5-08 · randomised placebo-controlled trial
   However, anaemia prevalence was significantly lower only in the FB group (1·0%) compared with the C group (10·4%, P = 0·006), with the SUP group being intermediate (7·4%).
   Br J Nutr, 2012 · checked 2026-09-22
7. coo-r5-01 · outbreak investigation with matched case-control study
   In the case-control study, 33 of 35 case patients (94%) consumed ready-to-bake commercial prepackaged cookie dough, compared with 4 of 36 controls (11%) (matched odds ratio = 41.3; P < .001); no other reported exposures were significantly associated with illness.
   Clin Infect Dis, 2012 · checked 2026-09-22
8. coo-r5-02 · national probability survey
   Thirty-five percent of consumers reported having tasted or eaten something with uncooked flour in it in the previous 12 months.
   J Food Prot, 2022 · checked 2026-09-22
9. coo-r5-04 · prospective cohort study
   Each 10μg/day increase in acrylamide intake was associated with an odds-ratio for SGA of 1.11 (95% Confidence Interval: 1.03,1.21), birth length change of -0.05cm (95% CI: -0.11,0.00) and birth weight change of -9.8g (95% CI: -21.3,1.7).
   Environ Res, 2016 · checked 2026-09-22