Cocoa contaminants, serving size and the limits of product tests

Updated 10 Oct 2026

Cocoa is an optional food I was considering, rather than a confirmed daily intake. The cocoa, tea and coffee investigation covers human research on potential benefits. Contaminant testing asks a separate question: what is in a particular product, in what amount, and how often would I consume it?

There is no need to turn an interesting food into a large daily supplement habit before answering those questions. Detecting a trace metal does not by itself mean a food should be avoided. Equally, the fact that a food contains beneficial compounds does not erase its contaminant exposure.

What original testing establishes

An FDA-authored 2018 study measured lead and cadmium in a convenience sample of cocoa powder, chocolate and cocoa nib products sold in the United States. Cadmium concentrations ranged from 0.004 to 3.15 mg/kg and lead from below detection to 0.38 mg/kg. Concentrations varied with product type, cocoa percentage and reported origin; those relationships do not identify the concentration of an untested brand. The inspected abstract supports the concentration ranges, not a current ranking of specific retail products. Original FDA analytical study.

A widely discussed 2024 analysis tested 72 cocoa-containing products in four cohorts collected during 2014–2022. Its headline lead result was 31 of 72 above California’s 0.5 µg/day reproductive-toxicity safe-harbor benchmark under its serving assumption. Its comparable cadmium result used a smaller denominator: 13 of 37, rather than 35% of all 72. Serving and daily-dose data were incomplete for some samples. Products were selected largely through consumer popularity surveys, and products with higher contamination were often excluded from subsequent testing. The apparent improvement across years therefore cannot be confidently interpreted as the whole market becoming cleaner. “Organic” did not provide a reliable contaminant guarantee in this dataset. Original 2024 paper, methods and tables.

This paper measured products; it did not measure clinical harm in people eating them. A percentage above one warning benchmark is not a percentage proven toxic, recalled, or above an FDA chocolate action level.

The newest FDA release includes cocoa powder

The FDA’s 2026 Total Diet Study release adds FY2021–FY2022 monitoring. Its two national cocoa-powder composite results provide a useful serving-size example:

The exposure column is calculated as concentration × 5 g, with 1,000 ppb equal to 1 µg/g. Five grams is an illustration, not a tested safe dose or my reported intake. Doubling the portion doubles that source’s exposure. The results are pooled national samples, not two named cocoa brands, and do not establish what is in Trader Joe’s cocoa today. FDA TDS interface, Elements Report, release date and collection years.

The national sampling design selects products representing substantial market share and may combine brands or varieties. That is useful for population monitoring; it can obscure a higher or lower individual product. A current product or lot assay answers a different question. FDA TDS sampling design.

Why different thresholds produce different headlines

California Proposition 65 lists a 0.5 µg/day lead reproductive-toxicity safe-harbor level and a 4.1 µg/day oral cadmium level. These determine aspects of warning requirements and their exemptions. They are not nutrient RDAs, desired intakes, or boundaries at which a person abruptly becomes poisoned. A product warning does not disclose the actual measured dose. OEHHA safe-harbor table.

FDA lead interim reference levels are 2.2 µg/day for children and 8.8 µg/day for females of childbearing age. They address exposure across food, with the latter protecting fetal and nursing-infant exposure. They are not a personal allowance for an adult man to spend on cocoa. FDA says no safe lead exposure level has been identified, while also explaining that detecting lead does not automatically mean abandoning a nutritious food. FDA lead references and interpretation.

FDA cadmium assessment uses a toxicological reference range of 0.21–0.36 µg/kg body weight/day, based on chronic bone and kidney endpoints. This is a range used in health assessment, not a cocoa-specific permitted dose. Cadmium exposure from other foods matters too. Older cocoa discussions that say FDA has no cadmium reference need updating. Current FDA cadmium assessment, original 2023 reference-value derivation.

The difference between these reference systems explains why the same serving can produce different headlines. It does not make any one system a guarantee of zero risk.

There is relevant counterevidence, with a different scope

A 2026 study analyzed 36 milk and dark chocolate samples from six brands, then modeled US consumption for different age groups. Its abstract reports no non-cancer concern under central consumption assumptions, while some high-end assumptions flagged possible lead concerns. Some arsenic risk calculations were also flagged, although arsenic was quantified in only two samples. The authors judged typical chocolate-bar exposure unlikely to cause harm under their model. Only the abstract was accessible for this review, so the detailed assumptions and product-specific results were not independently inspected. Original 2026 chocolate-bar study.

That is useful counterevidence to a claim that ordinary chocolate consumption is inherently dangerous. It does not establish the exposure from large daily amounts of concentrated powder, reproduce a particular brand’s current batch, or measure long-term outcomes in cocoa consumers.

A practical decision for an optional food

If I try cocoa, a small amount for enjoyment is a different commitment from making it a daily high-dose intervention. For frequent use, a useful testing report identifies the actual product, sampling date or lot, measured lead and cadmium concentrations, laboratory method, and the serving mass used for any exposure calculation. “Tested,” “organic,” “premium,” or “below limits” without those details cannot rank two products reliably.

A historic test of one bar does not permanently certify that brand’s powder or future lots. An undetected result means below that method’s detection or reporting limit, rather than exactly zero. Conversely, a recall or regulator finding concerning an identified product is actionable evidence and should not be diluted into a vague debate about all chocolate.

For now, cocoa remains optional. The useful next information would be the actual product and amount, if it becomes part of my routine. No current log establishes my cocoa exposure, a metal-related illness, or a reason to replace foods I already enjoy.

Back to investigations