How We Built the Produce Aisle Cheat Sheet

This is the working behind the produce aisle cheat sheet — what's in it, how each commodity got placed, and what we left out.

Every number on the sheet comes from the U.S. Department of Agriculture's Pesticide Data Program (PDP) — the federal government's own residue-testing program, running since 1991. We did not estimate, average, or infer anything. Where PDP hadn't tested a commodity recently enough for us to say something responsible about it, we left it off the sheet rather than guess.

The most important thing to know before you read any of this: all produce is better than no produce. Whether you buy organic or conventional, eating more fruits and vegetables is one of the best things you can do for your family's health. Across the program as a whole, more than 99 percent of samples come in under EPA's legal limit; the rate varies by commodity, and every row on the sheet states its own. The American Academy of Pediatrics is direct about the choice: families "should not avoid consuming produce just because organics are not available or possible for the family to purchase." The sheet exists for one narrower question: if your grocery budget allows a little organic spending, where does it do the most?

Why we build from USDA data rather than a shopping list

Several nonprofit organizations publish their own branded produce shopping lists. We don't use them, for two reasons.

Licensing. Those guides are typically copyrighted, and their specific rankings and scoring methodology aren't ours to republish or imitate. We only build data tables from sources with clear public-domain or redistribution rights.

We can build our own, and it's more transparent. PDP publishes the underlying sample-level data for free, every year, for anyone to check. Building our own placement logic from that primary data — and showing our work — means you don't have to take a third party's word for a "score." You can click through to the same government report we did and check the same numbers we checked.

USDA's Pesticide Data Program is a public federal dataset used by many organizations that analyze pesticide residues in food. Our methodology is independently defined: we use PDP results directly and apply the thresholds below.

What PDP actually measures

PDP is a voluntary federal program, run by USDA's Agricultural Marketing Service, that buys real produce off real shelves across nine states and tests it in a lab — the same thing you'd buy at a grocery store, washed the way a normal shopper would wash it, about 15 to 20 seconds under running water with no special soap. It reports results two ways that matter here:

• Whether a residue was detected at all, and how many different pesticides showed up in the same sample.
• Whether any residue exceeded the tolerance — the legal maximum EPA allows, set with a wide built-in safety margin. Exceeding it is rare, and gets reported to the FDA.

Detection is not the same as danger. A residue can show up at a tiny fraction of the legal limit, and most do. A commodity can show residue on nearly every sample and still record zero tolerance exceedances, because the levels found are simply low. Potatoes are the clearest example in our data: 98 percent of tested samples had a detectable residue — almost always a single postharvest sprout-inhibitor — and zero exceedances across every cycle we checked. Potatoes are in Column B.

That's exactly why we don't sort on raw detection percentage.

How we sorted the two columns

We used two numbers PDP itself publishes, and only those two.

1. Tolerance exceedance rate — the share of a commodity's samples reported to FDA as exceeding the EPA limit. The one number in PDP's data that reflects an actual regulatory finding rather than "residue present at any level."
2. Multi-residue load — the share of samples carrying four or more different pesticides at once, from PDP's own "Number of Pesticides Detected per Sample" appendix. This is not a hazard measurement; a food can carry several individually-approved pesticides and still sit within every tolerance. It's a proxy for how many distinct chemical treatments went into that growing season, which is what buying organic actually reduces.

A commodity lands in Column A — organic first, if you can — if the most recent PDP cycle testing it showed an exceedance rate at or above 1 percent of samples, or an exceedance recorded in two or more separate cycles (even at low rates), or 40 percent or more of samples carrying four-plus simultaneous residues.

Everything else is Column B — conventional is fine. Residues, where found, tended to be one or two familiar compounds at low levels, and outright exceedances were rare to absent across the cycles we checked.

Apples are the clearest illustration of why the rule is written this way. Every 2024 apple sample cleared the legal limit — zero exceedances — but 58.5 percent carried four or more different residues at once, well over the third trigger. Apples are in Column A on multi-residue load alone, and that is a statement about how the crop is grown, not a warning about eating one.

The two close calls

Applying a rule you wrote in advance sometimes produces an answer you didn't expect. Both times it did, we followed the rule and said so here rather than quietly overriding it.

Celery moved to Column A. Its 2021 and 2023 cycles each recorded exceedances — 0.56 percent and 0.86 percent. Neither crosses the 1 percent line on its own, and our first draft placed celery in Column B for that reason. But the second trigger doesn't require any single cycle to cross anything alone; it asks whether an exceedance was recorded in two or more separate cycles, and celery's plainly was. Reading the rule as written moves it to Column A.

Winter squash stayed in Column B. Its exceedance rate is 0.99 percent — under the 1 percent line, by one hundredth of a percentage point. Nothing about that gap is meaningful in itself, and we could have rounded it into Column A. We didn't, because a threshold you set aside when it's inconvenient isn't a threshold. It's marked on the sheet as a close call instead.

Both rows carry a "borderline" marker on the sheet, so you can see where the rule was doing real work rather than confirming something obvious.

PDP rotates what it tests, and that matters

PDP does not test the same twenty or so fruits and vegetables every year. It rotates through a much longer list on a multi-year cycle. That has one real consequence: a commodity's report year may be older than this page, and that doesn't mean it has gone unchecked or gotten worse — it means PDP simply hasn't re-sampled it yet. We show the exact year next to every commodity so you can see this for yourself. Green beans and sweet bell peppers are the ones to watch: neither has been retested since 2022 and 2021 respectively, despite showing some of the clearest Column A signals in our data.

Similarly, a commodity's absence from the sheet does not mean it's clean. It usually means PDP hasn't sampled it in the cycles we reviewed, or is testing it in a round that hasn't published yet. Spinach, strawberries, kale, sweet potatoes, brussels sprouts, bulb onions, snap peas, milk and oats all fall in PDP's 2025 sampling round, which had not published when we built this. They are deliberately left off rather than guessed at, and will be added when that data is public.

The cycles we used

We pulled the four most recent published PDP Annual Summaries — 2024 (the latest available), 2023, 2022 and 2021. You can browse every year PDP has published, back to 1992, and download the full raw sample-level datasets yourself from USDA's own Pesticide Data Program page.

What this page is not

• It is not a hazard ranking. It does not tell you conventional produce is unsafe. The government's own data says the large majority of samples of every commodity tested, in both columns, comes in under the legal safety limit.
• It is not a reason to eat less produce. The health cost of a family eating less produce because organic is unaffordable is real, and larger than any residue difference this sheet describes.
• It is not static. PDP republishes annually and rotates commodities. We re-run this research when a new cycle publishes, and the sheet carries the years it was built from.