In 2025 a study set out to test whether the best known consumer list about pesticides actually holds up. The researchers built a pesticide load index covering 44 kinds of produce, scored the diets of 1837 Americans against it, and compared the scores with what was measurable in those people’s urine. Nothing lined up: the coefficient came out at 0.02, with a confidence interval running from -0.34 to 0.38. An association appeared only after potatoes were dropped from the calculation.
The authors are worth noting. They work for the Environmental Working Group, the organization that publishes the Dirty Dozen every year. They were testing their own instrument, and they published a result that does it no favors.
What the list is
Each year EWG takes data from the US federal pesticide residue testing program, ranks 46 popular fruits and vegetables by contamination, and labels the top twelve the Dirty Dozen. The other end of the ranking becomes the Clean Fifteen. The list then travels through the news, and shoppers buy strawberries from the organic shelf and cabbage from the ordinary one.
The core objection from risk assessment specialists runs like this: consumer guides of this kind borrow some hazard assessment principles without performing a dietary risk assessment. The list answers the question of where residues are found more often, and the shopper reads it as an answer to which food is more dangerous. Those are different questions, and dose sits between them.
What the independent calculations found
The most thorough check still belongs to the University of California, Davis. Winter and Katz took the ten most frequently detected residues on each of the twelve listed commodities and modeled exposure probabilistically. That produced 120 estimates.
| Result across 120 exposure estimates | Count |
|---|---|
| Above 1% of the reference dose | 1 (methamidophos on bell peppers, 2%) |
| Above 0.1% of the reference dose | 7 (5.8%) |
| Below 0.01% of the reference dose | three quarters |
| Below 0.001% of the reference dose | 40.8% |
Their three conclusions: exposure to the most commonly detected residues on these commodities poses negligible risk, substituting organic versions produces no appreciable reduction in consumer risk, and the ranking methodology itself lacks scientific credibility. What organic farming permits in its own right, and what that costs soil and pollinators, is covered separately.
A 2024 paper repeated the exercise on a more recent edition of the list, taking the most frequently detected pesticide for each item and comparing estimated daily intake against US Environmental Protection Agency guidance values. Every combination came in below the benchmark in every scenario, children included, and the quantity of produce needed to reach the limit far exceeds what people eat. It is worth knowing who ran the numbers: the authors work at risk assessment consultancies. There is no conflict of interest declaration in the PubMed record, so this is a fact about affiliation rather than about motive, and you can weigh it yourself. The Davis work carries no such question, and it reaches the same conclusion.
How much turns up in Europe
An American list describes an American market. The European numbers come from EFSA, and for 2022 they read as follows: of 110,829 samples analyzed, 96.3 percent fell below the maximum residue level, 3.7 percent exceeded it, and 2.2 percent were non-compliant once measurement uncertainty was accounted for. In the coordinated program subset of 11,727 samples, 0.9 percent were non-compliant. Veterinary drugs are counted the same way, and the Danish 20,000 pork samples a year return zero to two over the limit. The overall assessed risk to EU consumer health is described as low. Worth reading beside any such list is glyphosate in grains, where the deciding factor is spray timing rather than the crop.
Those 2.2 percent deserve separate attention. They represent violations on specific consignments rather than a background level across the market, and they are dealt with by enforcement rather than by which shelf you buy from.
Where the criticism misses
Every check described above counts one substance at a time and looks at chronic exposure. The authors of the 2024 paper say themselves that acute scenarios and cumulative risk still need assessing. That gap is real.
A Brazilian group showed in 2026 what it looks like in practice. Eleven pesticides across 20 plant-based foods, 60 samples. Imidacloprid in zucchini exceeded its limit by more than 70-fold. Cumulative non-carcinogenic risk put cassava, strawberry and spinach at the top, with a hazard index of 3.16 for cassava, and a single compound, fipronil, accounted for 83 to 99 percent of that total. The authors put it plainly: regulatory compliance on its own does not capture the health risks of mixtures.
That is the Brazilian market, with a different set of approved substances, and the figures do not transfer to a European shelf. The principle does. “Each substance is within limits” and “the mixture is within limits” are two different statements, and nobody checks the second one systematically.
What EWG found when it tested itself
Back to the study this article opened with. The indices were built three ways, using different indicators of contamination and toxicity, and none of them tracked the biomarkers. Restricting the calculation to pesticides with a matching urinary marker in NHANES raised the association slightly, and it still fell short of significance.
Dropping potatoes changed the picture: produce consumption weighted by contamination then did line up with higher urinary markers for organophosphate, pyrethroid and neonicotinoid insecticides. Why potatoes have that effect the abstract does not explain; the authors simply recommend excluding them from future calculations.
The honest reading is that ranking produce by contamination captures something, weakly and unstably, and the whole effect disappears without one crop. For a scientific instrument that is a poor showing. For a list people use weekly to decide what goes in the basket, it is worse.
What to do in the kitchen
The practical answer is dull and outperforms the list. Wash fruit and vegetables under running water: surface residue comes off mechanically, and this is the cheapest lever available. What a baking soda soak removes, and why it is not universal, is covered separately. Anything dissolved inside the tissue is beyond washing altogether, and nitrate shows that most clearly.
Second, variety matters more than picking a “clean” item. The same pesticide on the same produce every day is the worst case available, and changing what you eat breaks it easily. A list of E numbers on a label works the same way, sorting substances by register rather than by risk, which we take apart in our piece on food additives.
Third, the money. The Davis authors wrote back in 2011 that swapping these twelve items for organic versions gives the consumer no appreciable reduction in risk. What to do with that is a personal call, though a trade in which a pricier shelf means fewer vegetables in the diet looks like a poor one.
Key points
- The list ranks produce by how often residues are detected, not by dose. Those are different things, and shoppers read it in the sense it was not written in.
- Of 120 exposure estimates in the Davis work, one exceeded 1 percent of the reference dose and three quarters fell below 0.01 percent.
- In Europe in 2022, 96.3 percent of 110,829 samples were below the residue limit and 2.2 percent were non-compliant.
- The weak spot in all these checks is genuine: they count one substance at a time and chronic exposure only. Cumulative mixture risk goes unassessed, and the Brazilian data show why that matters.
- EWG’s own test found no association between its contamination index and urinary biomarkers until potatoes were removed from the calculation.
- The working levers in a kitchen are running water and a varied diet. Swapping twelve items for organic versions gives no appreciable reduction in risk.
How washing works and where it stops is covered in our piece on wash water sanitizers. The wider picture of fresh produce risk sits in the food safety hub.
