The Dirty Dozen and Clean Fifteen, explained

Every spring, the Environmental Working Group publishes two lists: a “Dirty Dozen” of produce showing the most pesticide residue in government testing, and a “Clean Fifteen” showing the least. They are among the most-read pieces of food content published each year, and they generate an equally reliable annual round of criticism.

Both the lists and the criticism are worth understanding, because they are arguing about different things.

What the lists get right

The underlying data is real. It comes from USDA and FDA monitoring programs that test thousands of samples a year using validated methods. And the ranking they produce is broadly reproducible: build your own index from detection frequency and residues per sample and you will land on a similar ordering.

Our own residue evidence index — built independently from detection frequency, mean residues per sample, and proximity to EPA benchmarks — puts spinach, grapes, strawberries, kale, peaches, and celery at the top, and avocados, sweet corn, onions, asparagus, and pineapple at the bottom. That overlap is not coincidence. It reflects agronomy: thin edible skins and large leaf surfaces collect more residue than husks and thick peels do.

What the criticism gets right

The lists rank on detection metrics rather than on dose relative to safety benchmarks. A sample can count toward a commodity’s ranking at a level thousands of times below the tolerance the EPA sets for that compound. Across the whole monitoring program, more than 99% of samples come in below tolerance.

Food-science researchers have made the reasonable argument that presenting a detection ranking as a shopping warning implies a risk gradient the data does not establish, and that discouraging produce consumption carries a well-documented health cost while the residues in question do not.

Why we publish a ranking anyway

Because there is a real question underneath, and it is not a safety question. It is a budgeting question: given that you will spend a limited amount on organic, where does that money buy the most measurable change?

That question needs a ranking. It does not need the ranking to be a risk warning, and we take some care that ours is not framed as one. The scores on this site are called relative residue evidence and organic upgrade priority. Not danger, not toxicity, not safety.

Where our ranking differs

Two methodological choices produce a different order in places.

We weight by how often you eat the food. A large gap on a fruit bought twice a summer matters less than a moderate gap on something eaten five days a week. This is why apples rank above nectarines on our foods worth buying organic list despite indexing lower on raw residue evidence.

We refuse to rank thin data. Below 50 conventional samples we decline. Where a commodity has no usable organic sample set — blackberries, for instance — the verdict is “insufficient data” rather than an inferred one. Annual lists generally rank everything.

We also show price. A ranking that ignores what the upgrade costs cannot tell you whether to make it, and the organic premium ranges from about 50 cents a pound on bananas to $2.50 on cherries.

What to do at the shelf

1. If a food is on the higher-residue end and you eat it weekly and the organic version is within a couple of dollars — buy organic.
2. If it is on the lower-residue end, buy conventional and put the money elsewhere. Avocados, corn, onions, and pineapple are the clearest cases.
3. If the organic version is out of stock or out of budget, buy the conventional one anyway. Not buying produce is the worse outcome, and it is not a close call.

The versions to read

The annual lists are worth reading for the year’s specific findings. Our version is here:

• Fruits with higher residue evidence
• Vegetables with higher residue evidence
• Lower-residue fruits and vegetables
• Foods worth buying organic, weighted by how often you eat them

Then price the upgrades against your own budget, which is the step no annual list can do for you.

Last reviewed 2026-05.