Where lead, nickel, mould toxins, combustion products, acrylamide and pesticide residues enter cocoa, and what happens to them on the way to chocolate.
18 studies2005 to 2026Each checked against its published abstract
What these studies add up to
Lead is a surface contaminant. Isotope work places it on the shell, acquired after harvest, and found no measurable addition in transport or manufacture.
Ochratoxin A is in the shell too. Three studies and an international code agree, and removing the shell took out nine-tenths or more.
So is at least one insecticide, found more often and at higher levels in the shell than in the shelled bean in Ghana.
Roasting makes some contaminants: acrylamide rose two to three fold, and light polycyclic aromatic hydrocarbons formed, though not the heavy ones the law regulates.
Sensitive methods find traces almost everywhere. Ochratoxin A was in every Canadian sample and aflatoxins in half of Spanish cocoa powders, at levels of a few parts per billion or less.
The benchmark decides the story. The same 72 American products were 43% over one lead level and 97% under another.
Shelling by hand removed more than 95% of the ochratoxin A from most samples
Does removing the shell change the ochratoxin A content of cocoa beans?
How
Twenty-two cocoa samples analysed for ochratoxin A by HPLC with fluorescence detection before (whole bean) and after manual shelling (nib).
Finding
Fourteen of the twenty-two samples (64%) lost more than 95% of their ochratoxin A through shelling, six lost 65-95%, and only one showed a reduction of less than 50%. The authors conclude that ochratoxin A contamination in cocoa beans is concentrated in the shell.
Does not establish: It does not show that finished cocoa products are free of ochratoxin A; surveys of cocoa powder and chocolate still detect it.
Limits: Manual shelling of a small number of samples; industrial winnowing leaves some shell in the nib.
What effect do the usual industrial treatments have on ochratoxin A in cocoa?
How
Cocoa pods from Côte d'Ivoire prepared under conditions promoting fungal growth, then put through roasting, shelling, crushing, pressing and additive addition, with ochratoxin A measured by HPLC at each stage.
Finding
Ochratoxin A in the artificially contaminated beans ranged from 3.4 to 44.7 micrograms per kg with a mean of 22.9 micrograms per kg, and was mainly concentrated in the shell (93%). Roasting, shelling and additive addition decreased levels by 24-40, 76 and 52% respectively, for an overall reduction of approximately 91%.
Does not establish: It does not describe naturally occurring levels in traded beans, which other surveys report as much lower.
Limits: Artificially high contamination; the abstract gives no sample number.
Can lead to neodymium ratios and isotope ratios distinguish soil-derived lead from contamination in cocoa, and show where it enters?
How
Measurement of lead, neodymium and samarium concentrations and of lead and strontium isotope ratios in cocoa shells and cocoa powders. The abstract gives no sample number.
Finding
Shells contain neodymium and samarium as well as lead, apparently taken up more slowly than lead when beans are dried on bare soil. Samples plotting above the derived lead to neodymium curves are judged probably contaminated. The analysis pointed to African samples being uncontaminated and samples from Asia containing 50% anthropogenic lead of Australian origin. No measurable lead contamination occurred during transport of beans and manufacture of chocolate.
Does not establish: It does not support the idea that lead is added in the factory. Nor does it quantify how much shell-borne lead reaches finished chocolate.
Limits: A single-author geochemical study; sample numbers are not stated in the abstract and the contamination test rests on fitted curves.
What levels of ochratoxin A occur in cocoa and chocolate available in Canada?
How
32 samples of cocoa powder (16 alkalized and 16 natural) and 28 chocolate samples (21 dark or baking chocolate and seven milk chocolate) analysed by liquid chromatography with fluorescence detection after immunoaffinity clean-up.
Finding
All 32 cocoa powders contained ochratoxin A, at 0.25 to 7.8 ng/g; six exceeded 2 ng/g. Detection in the 28 chocolates was also 100%, at 0.05 to 1.4 ng/g; one sample was above 1 ng/g. The 2 and 1 ng/g figures were limits the European Union had previously considered for cocoa and chocolate.
Does not establish: Detection in every sample does not mean a health concern was shown: the abstract reports concentrations only and makes no exposure assessment.
Limits: Sixty samples from one country at one period.
Do aflatoxin-producing fungi and aflatoxins occur in cocoa on Brazilian farms?
How
226 cocoa samples collected on Brazilian farms at various stages of fermentation, drying and storage; fungi isolated and identified, and aflatoxins determined by HPLC after immunoaffinity clean-up.
Finding
A total of 819 potentially aflatoxigenic fungi were isolated (Aspergillus flavus, A. parasiticus and A. nomius), with a considerable increase in numbers during drying and storage. Despite this prevalence, only low levels of aflatoxin were found in the cocoa samples, which the authors take to suggest limiting factors to aflatoxin accumulation in the beans.
Does not establish: It does not show that cocoa is free of aflatoxins: a later Brazilian survey of 123 bean samples reported a mean of 5.7 micrograms per kg total aflatoxins, and a 2026 Spanish retail survey detected aflatoxins in about half of cocoa powders.
Limits: Farm samples from Brazil; the abstract gives no concentrations.
How is ochratoxin A distributed among cocoa processing fractions, and how much is lost in making chocolate?
How
168 samples from fractions obtained during cocoa processing (shell, nibs, liquor, butter, cake and cocoa powder) analysed by immunoaffinity clean-up and HPLC, plus a chocolate manufacturing experiment.
Finding
The highest levels of ochratoxin A were found in the shell, cocoa powder and cocoa cake. Cocoa butter was the least contaminated. Under the conditions used and at the contamination level present, 93.6% of the ochratoxin A in the beans was reduced during chocolate production.
Does not establish: It does not show that roasting destroys ochratoxin A, and it does not apply to cocoa powder, where the toxin concentrates.
Limits: One manufacturing experiment at one contamination level; the reduction includes dilution by sugar and other ingredients as well as shell removal.
Are neonicotinoid insecticide residues present in fermented and dried cocoa beans from Ghana, and how are they distributed between shell and bean?
How
Fermented and dried beans from major cocoa-producing regions of Ghana, with shells removed and analysed separately, using a QuEChERS extraction and LC-MS/MS. The abstract gives no sample number.
Finding
Imidacloprid was the most frequently found neonicotinoid, quantified in more than 10% of deshelled bean samples and 30% of shell samples. Concentrations ranged from 11.5 to 35.6 micrograms per kg in beans and 11.8 to 214 micrograms per kg in shells. Limits of quantification ranged from 3 to 10 micrograms per kg.
Does not establish: It does not show that residues exceeded a maximum residue limit; the authors' reference to unsafe levels is a caution about build-up, not a reported exceedance.
Limits: One country; sample number not stated in the abstract; no comparison with a named legal limit.
What levels of organochlorine pesticides are in dried cocoa beans from produce stores in two towns of south-western Nigeria?
How
Dried cocoa beans from produce stores at Ondo and Ile-Ife were extracted with dichloromethane, cleaned up on silica gel and analysed by gas chromatography with electron capture detection. The abstract does not give the number of samples.
Finding
Organochlorines were present in beans from both towns. Mean levels of p,p'-DDT were 82.17 ± 54.53 ng/g at Ondo and 57.76 ± 81.48 ng/g at Ile-Ife. The authors report that the levels were within maximum residue limits set by the World Health Organization and Food and Agriculture Organization.
Does not establish: It does not show how common such residues are in Nigerian cocoa generally, whether the DDT was recently applied or a legacy in soil, or what reaches chocolate.
Limits: Two towns, an unstated number of samples, and standard deviations as large as the means. Which limits were used for comparison is not specified in the abstract.
How do nickel, cadmium and lead levels change from raw cocoa through the steps of chocolate manufacture?
How
Measurement of cadmium, lead and nickel in raw cocoa and in the masses from successive manufacturing steps in three different companies. The abstract gives no sample number.
Finding
Total reductions varied by production line: 10.5–33% for cadmium, 0–100% for lead and 11–42% for nickel. The two steps with the greatest effect were winnowing of the cocoa bean shell and conching. Nickel was the most abundant of the three, at a maximum of 12.1 mg/kg in raw cocoa and 4.5 mg/kg in chocolate. Chocolate from one producer contained 0.43 mg/kg cadmium.
Does not establish: The abstract's statement that nickel poses a serious risk to children's health is the authors' judgement; the abstract reports no exposure calculation to support it.
Limits: Three factories and their raw materials; reductions reflect dilution by other ingredients as well as removal.
What acrylamide levels occur in cocoa and chocolate products?
How
Survey of 94 semi-finished cocoa products (beans, nibs, powder, masses and cocoa butter) and 269 cocoa and chocolate products sourced from the German market.
Finding
Acrylamide ranged from below the limit of quantification (less than 30 micrograms per kg) to 490 micrograms per kg. Semi-finished cocoa products had a higher mean (190 micrograms per kg) than cocoa and chocolate products (50 micrograms per kg). All samples were distinctly below the general German signal value of 1000 micrograms per kg.
Does not establish: It does not show that these levels are without risk. The comparison is with a general signal value; the abstract notes that no specific signal or EU benchmark value existed for cocoa and chocolate.
Limits: One national market at one period; the abstract does not say how products were selected.
How do PAH levels differ between cocoa beans of different origins and change through one processing line?
How
Cocoa beans of several varieties from different countries and their derived products from one technological line, analysed for 19 PAHs by HPLC and confirmed by GC-MS. The abstract gives no sample number.
Finding
Raw beans of different varieties and origins differed significantly in total PAH content. The highest sums of 19 PAHs were in roasted beans, cocoa mass and cocoa butter (16.69-74.15 micrograms per kg of fat). Roasting at 160 °C led to PAH formation, though not of the heavy compounds; lowering the temperature to 140 °C while extending the time reduced total contamination only slightly. Light PAHs predominated and the sum of the 4 heavy marker PAHs was much lower than the legal maximum.
Does not establish: It does not examine beans dried over wood fires or with direct flue gases, which is the route by which heavy PAHs are usually said to reach cocoa, so it does not show that smoke-dried cocoa is low in PAHs.
Limits: One processing line and a single-author study; origins are not named in the abstract.
Which raw materials and process steps determine furan and acrylamide levels in dark chocolate?
How
Furan and acrylamide measured by GC-MS methods in cocoa and non-cocoa raw materials, in masses from processing stages and in chocolates from three factories. The abstract gives no sample number.
Finding
Among raw materials, cocoa masses and powders held the most acrylamide (83.0-127.5 ng/g). Roasting of cocoa beans increased acrylamide 2-3-fold. The results indicate acrylamide might be formed during wet conching. The finished chocolates contained 61.8-108.0 ng/g acrylamide.
Does not establish: It does not test alkalisation, so it says nothing on whether Dutch-processing changes acrylamide.
Limits: Three factories; the abstract gives no roasting temperatures.
Can a combined quantitative and screening method be used for routine pesticide residue analysis of cocoa beans, and what does it find?
How
A validated multiresidue method (quantitative for 15 pesticides, screening for 110) applied to 137 imported and domestic dried cocoa bean samples.
Finding
32% of the samples were positive for ametryn, chlorpyrifos, isoprocarb or metalaxyl. The abstract gives no concentrations and does not say whether any maximum residue limit was exceeded.
Does not establish: A positive finding is a detection, not a breach of a legal limit, and the abstract does not report any breach.
Limits: Primarily a method paper; the abstract does not describe how samples were chosen or their origins.
What do smallholder cocoa farmers in Ghana know about pesticides, and how do they apply, store and dispose of them?
How
A questionnaire survey of four hundred and four cocoa farmers chosen randomly from 26 communities in four cocoa-growing regions of Ghana.
Finding
87% of respondents belonged to cooperatives and certification groups. About 70% of insecticides used were approved by the Ghana Cocoa Board, with neonicotinoids and pyrethroids the most used classes. Most farmers showed improper storage, application and disposal practices despite claiming adequate knowledge.
Does not establish: It does not show what residues result from these practices, or that the 30% of products outside the approved list left residues above any limit.
Limits: Self-reported practice. No residue was measured in any bean, and the symptoms farmers reported were not clinically assessed.
How often do US dark chocolate and cocoa products exceed benchmark doses for lead, cadmium and arsenic, and has this changed over time?
How
72 consumer cocoa-containing products purchased from 2014 to 2022 in four cohorts (2014, 2016, 2019, 2022) and analysed for lead, cadmium and arsenic; compared with California Proposition 65 maximum allowable dose levels of 0.5, 4.1 and 10 micrograms per day respectively.
Finding
43% of products exceeded the Proposition 65 level for lead, 35% for cadmium and 0% for arsenic, while 97.2% (70 of 72) fell below the US FDA interim reference level for lead. Median concentrations of each metal were below the Proposition 65 levels, which the authors read as a large effect of outlying products. Products labelled organic were significantly more likely to show higher cadmium and lead.
Does not establish: It does not show that eating these products causes harm. The Proposition 65 levels are warning thresholds, not safety limits, and the two benchmarks used give very different pictures of the same data.
Limits: 72 products chosen by a testing company over several years, not a random sample of the market; the comparison is per single serving.
Can a QuEChERS extraction with low-temperature partitioning determine multiclass pesticide residues in cocoa beans, a fatty and difficult matrix?
How
Method development and validation for 15 multiclass pesticide residues in cocoa beans by gas chromatography-mass spectrometry, then applied to 15 cocoa samples from Bahia and Pará, Brazil.
Finding
Limits of quantification ranged from 13.9 to 23.1 μg kg-1 and recoveries were within 70-120%. Permethrin was the only pesticide detected, occurring in five samples.
Does not establish: It is not a survey of Brazilian cocoa, and the abstract gives no concentrations for the permethrin found.
Limits: Fifteen samples and fifteen target pesticides. The method's quantification limits are above the European default limit of 10 micrograms per kilogram, so it could not confirm compliance with that default.
Does chocolate bar consumption in the US contribute substantially to variation in blood lead levels?
How
Modelling study using the U.S. EPA All-Ages Lead Model with deterministic estimates and probabilistic (Monte Carlo) sensitivity analyses and age-specific exposure factors. No new blood measurements.
Finding
In central tendency scenarios chocolate contributed approximately 0.3-0.8 % of total lead intake and 0.5-0.7 % of lead in blood. In plausible high-end scenarios (approximately 50-75 % of a full-size bar daily) chocolate contributed approximately 3.1-6.5 % of blood lead levels. Soil and dust contributed 37 % and 51 % respectively to variance in young children's blood lead, and chocolate approximately 0.1 %.
Does not establish: It does not show that lead in chocolate is absent or irrelevant for a child who eats a high-lead product regularly; the high-end scenario still gave a contribution of several per cent.
Limits: A model whose output depends on the lead concentrations and consumption figures fed into it; the abstract does not state the concentrations assumed or the funding source.
How often do aflatoxins occur in cocoa-derived products sold in Spain, and what exposure results?
How
394 samples (109 cocoa powder, 160 dark and 125 milk chocolate bars) sold in Spain, analysed for four aflatoxins by a validated HPLC method with fluorescence detection; risk characterised by margin of exposure.
Finding
Aflatoxins were detected in 51.4% of cocoa powder and 28.8% of chocolate bars; dark chocolate had higher positivity (39.4%) than milk chocolate (15.2%). AFB1 was found in 43.1% of cocoa powder (mean 0.30 micrograms per kg). The highest AFB1 and AFG1 concentrations, in cocoa powder, were up to 3.0 and 3.3 micrograms per kg. Margins of exposure below the ten-thousand benchmark were calculated for adolescents, children and toddlers at average cocoa powder consumption.
Does not establish: It does not show harm. A margin of exposure below the conventional benchmark signals a possible concern for a genotoxic carcinogen, not a measured effect, and the authors describe chocolate as a minor source.
Limits: One national market; detection limits were very low, so a high detection rate reflects analytical sensitivity as well as occurrence.
How much lead on the shell ends up in finished chocolate, and by what route.
Whether drying beans over fires contaminates them with heavy polycyclic aromatic hydrocarbons. It is the standard explanation and no primary study was found.
Where nickel in cocoa comes from and how it varies. One factory study is all that was read.
How often cocoa beans in trade carry ochratoxin A or pesticide residues above a limit. The surveys read do not say.
What any of this means for products made from cocoa shell, where three of these contaminants concentrate.
These are questions the studies read for this page leave unanswered. They are not a claim that nobody has answered them.
We would like to set two Google Analytics cookies to count page views, and to count them in our own cookie-free measurement as well. They do not identify you and nothing is shared with advertisers. Decline and the site works exactly the same — see cookies for what each one does.