How Peel scores
On this page
Scoring version: 2026.1 · Page updated: [PUBLICATION DATE] · Questions or corrections: [CONTACT EMAIL]
Peel gives fresh fruits and vegetables a score from 0 to 100. The score is our opinion, reached by applying the method on this page to public government data. We publish the method so you can see exactly how each number is built, and so anyone can check it.
1. What the score measures
Mostly, pesticide residues found on the produce in U.S. government testing, judged against a stricter standard than U.S. law, plus a small nutrition component. The breakdown screen shows every factor, the points it earned, the pesticides detected and how their levels compare with the limits.
Conventional and organic produce are scored separately.
2. The data
- Residue testing: the U.S. Department of Agriculture's Pesticide Data Program (PDP), which buys produce from U.S. stores and distribution centers and tests it for hundreds of pesticides. Samples include domestic and imported produce as sold in the U.S. Only fresh produce samples are used; frozen, canned and processed samples are excluded.
- Limits and approval status: the EU Pesticides Database (European Commission) and Health Canada's Pest Management Regulatory Agency. U.S. EPA tolerances are shown for reference.
- Toxicity: the EU's harmonised hazard classifications (CLP Regulation, Annex VI).
- Nutrition: USDA FoodData Central.
Peel does not use data from advocacy groups' produce rankings, and brands and growers cannot pay to change a score.
3. Benchmarks: which standard you compare against
A benchmark is the set of limits and approval rules the residues are judged against. You choose it in the app.
| Benchmark | Limits used | "Approved" means |
|---|---|---|
| EU (default) | EU maximum residue levels (MRLs). Where the EU sets no specific limit, its default of 0.01 mg/kg | Approved as an active substance in the EU |
| Canada | Canadian MRLs. Where Canada sets no specific limit, its general limit of 0.1 mg/kg | Registered in Canada (at least one currently, fully registered product contains it) |
| Strictest available | For each pesticide, the lowest limit among the available benchmarks | The most restrictive status among them |
The U.S. is never a benchmark, because Peel's purpose is to show how produce looks against stricter international standards. U.S. EPA tolerances are always shown alongside for reference. Other benchmarks (Codex, Japan) are listed as "coming soon" and are not used.
Limits are legal and trade thresholds, not safety thresholds. An MRL is the highest residue a government allows in food it regulates, often set well below toxicological concern and sometimes set at the lowest level labs can measure. A residue above an EU limit on produce sold in the U.S. is not illegal in the U.S., and it does not by itself mean the food is unsafe.
4. The five factors (2026.1)
Each factor measures the share of test samples with a problem. That makes a pesticide found once count less than one found in most samples.
| Factor | Up to | How points are earned |
|---|---|---|
| Not approved in the benchmark | 30 | 30 × (1 − share of samples with at least one residue of a pesticide not approved in the benchmark) |
| Levels above benchmark limits | 25 | 25 × (1 − share of samples with at least one residue above the benchmark's limit) |
| How often residues appear | 20 | 20 × (1 − share of samples with any detected residue) |
| Toxicity of what was found | 15 | 15 × (1 − toxicity load, capped at 1). The load adds, for each pesticide, the share of samples it was found in times its hazard weight |
| Nutrition | 10 | 10 × the item's nutrient-density percentile among scored produce (fiber, vitamin C, potassium and folate per 100 kcal) |
Hazard weights (from each pesticide's EU CLP classification):
| Tier | Weight | CLP classes |
|---|---|---|
| Very high | 1.0 | Acute toxicity category 1–2; carcinogenic, mutagenic or reproductive toxicant category 1A/1B |
| High | 0.6 | Acute toxicity category 3; suspected carcinogen, mutagen or reproductive toxicant (category 2); organ damage from repeated exposure (STOT RE 1) |
| Moderate | 0.3 | Acute toxicity category 4; STOT RE 2; STOT SE 1–2 |
| Low | 0.1 | Classified, but none of the above (for example only irritation or environmental hazards) |
| Not classified in Annex VI | 0.3 | No harmonised classification available; treated as moderate |
Missing factors. If a factor can't be computed (for example, the benchmark has no approval status for the pesticides found, or there's no nutrition data), it is left out and the other factors are scaled up so the maximum is still 100. A pesticide whose approval status is unknown is not counted as unapproved.
Rounding. The total is rounded to a whole number, and the factor points in the breakdown are rounded so they add up exactly to the score.
Legacy chemicals. Some residues come from pesticides banned in the U.S. decades ago that persist in soil (for example DDT or dieldrin). They count in full, like any other residue, and are labeled "Legacy chemical" so it's clear no one sprayed them recently.
Worked example (illustrative numbers, not a real item)
100 samples. 20 had a pesticide not approved in the EU; 10 had a residue above an EU limit; 60 had any residue. Pesticide A (high tier) appeared in 20% of samples and pesticide B (low tier) in 50%. Nutrition is at the 70th percentile.
| Factor | Calculation | Points |
|---|---|---|
| Not approved | 30 × (1 − 0.20) | 24.0 |
| Above limits | 25 × (1 − 0.10) | 22.5 |
| How often | 20 × (1 − 0.60) | 8.0 |
| Toxicity | 15 × (1 − (0.20 × 0.6 + 0.50 × 0.1)) = 15 × 0.83 | 12.45 |
| Nutrition | 10 × 0.70 | 7.0 |
| Score | 74 (Good) |
5. Score bands
| Score | Band |
|---|---|
| 80–100 | Excellent |
| 60–79 | Good |
| 40–59 | Fair |
| 20–39 | Poor |
| 0–19 | Avoid |
The bands summarize residues relative to the chosen benchmark. "Avoid" means the item compares poorly with the benchmark on residues. It is not a statement that the item is unsafe to eat.
6. How much data we need, and how recent
- Test window. USDA tests each type of produce in some years and not others. Peel uses the most recent span of up to 3 calendar years that holds at least 30 samples. It never pools distant surveys (for example, 2025 results are never combined with 2016 results).
- Minimum of 30 samples. With fewer than 30 samples, Peel shows "Not enough data" instead of a score.
- Test years are always shown. Every score lists the years of testing it is based on.
- Older data flag. If the latest test year is more than 5 years before the score was published, the score is marked as based on older data, with an explanation. We still show it because it's the most recent government testing available, and we update it automatically when USDA publishes new results.
- Updates. Peel checks its sources weekly. USDA publishes new PDP results about once a year.
7. Origin, packaged items and mixes
- National score by default. Scores cover all samples sold in the U.S., domestic and imported. If a package names a single country of origin and USDA tested at least 30 samples from that country in the window, Peel shows the origin-specific score; otherwise it says it's using the national score.
- Packaged produce (clamshells, bags, fresh-cut) is matched to its produce type through its barcode and ingredient list.
- Mixes are estimates. For a mix such as a spring mix, each ingredient's score is weighted by its position in the ingredient list (weights halve down the list, then are normalized), and the result is always labeled "Estimated", with the range from the lowest to the highest ingredient score and a row for each ingredient. Ingredients without enough data are left out; if those with data make up less than half the weight, the mix shows "Not enough data".
- Salad kits: only the fresh produce is scored ("Greens only").
- Not scored: frozen, canned, dried, juiced, cooked or seasoned products, and anything with added non-produce ingredients.
8. What the score is not
- Not a safety rating or a health verdict. Government agencies, including USDA, report that residues found in U.S. testing are generally below U.S. legal tolerances. Peel compares them with stricter standards to help you choose, not to say a food is dangerous.
- Not a test of your item. It summarizes government samples of that kind of produce, not a particular brand, store, farm or batch.
- Not a nutrition score. Nutrition is 10 of 100 points.
- Not medical or dietary advice. Eating fruits and vegetables is good for you whatever their score.
- Not affected by recalls, labor information, seasonality or commercial relationships. Those are shown separately, or not at all.
9. Shown next to the score, never in it
- Food-safety recalls from FDA records, in FDA's words, with a link. "Recalled" appears only when a recall lists the exact barcode you scanned; otherwise recalls involving that kind of produce are shown as history.
- Season labels from USDA Market News, only where data supports them.
- Labor practices (planned): U.S. Department of Labor country-level findings, worded as the source words them.
10. Changes and corrections
Each score shows its scoring version. When we change the method, we publish a new version of this page, describe what changed and why, and rescore every item. If you think a score or a data point is wrong, email [CONTACT EMAIL] with the item; we'll check it against the source and correct errors we find.
Version history
| Version | Date | Change |
|---|---|---|
| 2026.1 | [DATE] | First public method |