Acrylamide became a food safety story in 2002, when Swedish researchers found it in ordinary cooked foods. It forms when sugars and the amino acid asparagine react at high temperature, the same Maillard chemistry that browns toast, chips, biscuits and coffee. It is classified by the International Agency for Research on Cancer as a probable human carcinogen based on animal studies, which is why coffee ended up carrying warning labels in California. The research on coffee specifically is more interesting than the warning suggests, and the roast level finding is the opposite of what most people assume.
Where acrylamide comes from
Acrylamide is not added to anything and it is not a contaminant in the usual sense. It is created by cooking. Reducing sugars react with asparagine at temperatures above about 120 degrees Celsius, producing acrylamide as a by-product of the reactions that also produce colour and flavour. Green coffee beans contain both precursors, and roasting applies exactly the conditions required.
This matters for how to think about it. You cannot remove acrylamide from coffee without changing the process that makes coffee taste like coffee, which is the central difficulty in every mitigation strategy proposed.
Lighter roasts contain more
The most useful and least intuitive finding in this literature concerns roast level. Lantz and colleagues[1] tracked acrylamide through roasting, storage and brewing, and found that its content peaks early in roasting, reflecting simultaneous formation and destruction, so levels in the fully roasted product are a small fraction of the peak reached earlier. The main factors affecting the final level were the arabica to robusta ratio, with robusta giving higher levels, and the time and degree of roast, with both shorter and lighter roasting at the edges of the normal range giving higher levels.
Schouten and colleagues[2] confirmed the pattern systematically, roasting arabica and robusta green beans to five degrees from light to dark and measuring acrylamide alongside trigonelline, nicotinic acid, caffeoylquinic acids and antioxidant activity. Acrylamide and antioxidant activity both reached a maximum in the early roasting degrees and then decreased with prolonged heating. The thermal reduction was greater for acrylamide than for antioxidant activity, which was only slightly reduced because degradation was balanced by the formation of new antioxidant compounds.
So a dark roast has less acrylamide than a light roast, and loses proportionally less of its antioxidant activity than you might expect. That inverts the usual health hierarchy, which treats light roasts as the healthier option because they retain more chlorogenic acid, discussed in our article on coffee acidity.
Storage lowers it, brewing carries it
Two further findings from Lantz and colleagues[1] are practically relevant. Acrylamide in roasted coffee declines during ambient storage, following second-order reaction kinetics with an activation energy of 73 kilojoules per mole, meaning fresher coffee contains more acrylamide than the same coffee after some weeks on a shelf. And the acrylamide in roasted coffee is largely extracted into the brew and is stable within the usual time before drinking, so it does reach the cup.
Their conclusion about mitigation is refreshingly blunt. Because the four main factors that determine acrylamide, bean species, roast degree, roast time and storage, also substantially affect the sensory character of the brew, and because process changes must stay within taste profiles consumers accept, only small reductions are realistically achievable.
Is dietary acrylamide actually causing cancer in people?
This is the question the warning labels assume and the evidence has not settled. Virk-Baker and colleagues[3] systematically reviewed the epidemiological literature on dietary acrylamide and human cancer. The majority of studies reported no statistically significant association between dietary acrylamide intake and various cancers, with a few reporting increased risk for renal, endometrial and ovarian cancers. They emphasised that exposure assessment in these studies has been inadequate, leading to potential misclassification or underestimation, and called for better exposure measurement.
The mechanistic case has also been questioned. Eisenbrand[4] reviewed the genotoxicity evidence and argued that at doses down to average consumer exposure, DNA lesions appeared only sporadically without dose dependence and close to human background levels, that the relevant DNA adducts have low mutagenic potency, and that toxicogenomic studies do not support a genotoxic mode of action. The conclusion was that the evidence argues for a non-genotoxic mode of action for the tumours seen in rodents, which, if accepted, would mean a tolerable daily intake could be defined rather than treating any exposure as carrying risk.
That is a live scientific argument rather than a settled position, and it matters enormously, because regulatory treatment of a genotoxic carcinogen differs fundamentally from that of a compound with a threshold.
How large is the modelled risk?
Quantitative assessments give a sense of scale. Li and colleagues[5] estimated average dietary acrylamide exposure in China at about 0.153 micrograms per kilogram of body weight per day for males and 0.155 for females, producing modelled excess lifetime cancer risks of 7.81 and 7.92 per 100,000. They attributed roughly 23,688 disability-adjusted life years of cancer across the Chinese population in 2016 to dietary acrylamide, including breast, endometrial, ovarian and kidney cancers.
Those are modelled figures derived from animal-based risk factors rather than measured human outcomes, and they apply to total diet rather than coffee alone. In most national diets the major acrylamide sources are fried potato products and baked goods, with coffee contributing a share that varies by country and coffee culture.
How much is in a cup, relative to other food
Putting coffee in dietary context is what most coverage skips. Acrylamide concentrations reported for roasted coffee are generally lower per unit weight than those for fried potato products, and a cup of coffee is mostly water, so the amount delivered per serving is modest. Instant coffee typically measures higher per gram of powder than roast and ground, though the serving size is smaller, a point taken up in our article on instant coffee.
The comparison that matters is total daily intake across the diet, which is how every risk assessment here was constructed, including Li and colleagues[5]. Focusing on the coffee while ignoring the fried and baked portions of a diet is an odd way to reduce acrylamide exposure, and it is precisely what a warning label on a coffee cup encourages.
Can anything be done at the roastery?
Mitigation has been studied seriously, because European regulators expect food businesses to apply reduction measures. The approaches explored include using enzymes to remove asparagine before roasting, adjusting roast profiles, selecting green coffee with lower precursor levels, and treating beans with additives.
The verdict in Lantz and colleagues[1] remains the honest summary: because the same variables that drive acrylamide also drive flavour, and because any change has to stay inside what drinkers will accept, only small reductions are achievable in practice. A roaster cannot make acrylamide-free coffee that tastes like coffee, which is why no serious producer claims to.
What IARC actually says about coffee
It is worth separating the compound from the drink. Acrylamide is classified as probably carcinogenic to humans, group 2A, on animal evidence. Coffee itself was evaluated separately, and in 2016 the IARC working group reported by Loomis and colleagues[6] moved coffee out of group 2B, finding the evidence inadequate to classify coffee as carcinogenic, while identifying very hot beverages above 65 degrees Celsius as probably carcinogenic in relation to oesophageal cancer.
That is a meaningful distinction. The same body that considers acrylamide probably carcinogenic declined to classify coffee as carcinogenic, having looked at coffee drinkers rather than at compounds in isolation. The wider picture is covered in our article on coffee and cancer risk, and coffee's association with lower mortality in our article on coffee and lifespan.
The California warning
California's Proposition 65 required warnings on coffee for years because acrylamide is on the state's list of chemicals known to cause cancer, and the listing applies to the chemical regardless of the food it appears in. In 2018 the state's coffee industry challenged this, and the regulatory position subsequently shifted to exempt coffee, on the reasoning that drinking coffee does not pose a significant cancer risk.
The episode is a good illustration of how a chemical-by-chemical approach can produce a conclusion that contradicts the evidence on the food itself. A drink whose consumption is associated with lower all-cause mortality was carrying a cancer warning because of one compound it contains.
Reading this from the Philippines
There is no local acrylamide warning scheme, and the relevant exposure question here is the diet as a whole rather than coffee in isolation. Fried and heavily browned starchy foods contribute more acrylamide than coffee does in most dietary assessments, and a Filipino diet built around rice, which is boiled rather than browned, has a different acrylamide profile from a Western one built around bread and fried potatoes.
The one coffee-specific note worth making is about robusta, which Lantz and colleagues[1] found gives higher acrylamide than arabica. Robusta is widely grown and drunk in the Philippines, including in traditional kapeng barako preparations, covered in our guide to barako. Nothing in this evidence suggests that is a reason to avoid it. It is a reason not to assume that figures measured on European arabica blends describe the local cup.
Why coffee keeps producing this kind of story
Acrylamide is the third contaminant story attached to coffee in recent decades, after mycotoxins and before whatever comes next, and they share a structure. A compound with genuine toxicological interest is detected in coffee by increasingly sensitive methods. The detection is reported as presence rather than as concentration against a threshold. A product category forms around avoiding it. And the epidemiology on people who drink coffee, which is where the question would actually be settled, points in a different direction entirely.
Holding that pattern in mind is more useful than memorising any number here. The measurements in this article are real and the chemistry is well understood. What remains unresolved is the step from measurement to harm, and that step is where both the warning labels and the premium products get ahead of the evidence.
What would settle it
The question that matters, whether dietary acrylamide at ordinary intakes causes cancer in humans, has a clear methodological bottleneck, and Virk-Baker and colleagues[3] named it: exposure assessment. Asking people what they ate and converting that into milligrams of a compound formed variably during cooking produces estimates with large errors, and those errors bias associations toward nothing.
The fix is biomarkers. Acrylamide forms measurable adducts with haemoglobin that reflect exposure over the preceding months, far more accurately than any questionnaire, and studies using them are the ones worth watching. Until enough of those accumulate, the honest position on coffee and acrylamide is the one this article takes: a real compound, at small doses, with an unresolved question attached.
The limits of this evidence
The roasting and storage studies are laboratory chemistry, done on specific beans under specific conditions, and commercial roasting varies. The epidemiology relies on estimating acrylamide intake from food frequency questionnaires, which the reviewers themselves identified as the field's central weakness. Animal carcinogenicity studies used doses far higher than dietary exposure, and whether the rodent tumour mechanisms apply to humans is the open question in the field. The modelled cancer risks are derived from those animal studies rather than observed in people.
The practical answer
Coffee contains acrylamide, formed during roasting, and the amount is small, declines with darker roasting and with storage, and is largely extracted into the brew. Whether dietary acrylamide at these levels causes cancer in humans is unresolved: most epidemiological studies find no significant association, exposure measurement is poor, and the mechanism is disputed. The same agency that classifies acrylamide as probably carcinogenic declined to classify coffee as carcinogenic after reviewing coffee drinkers. If you want to reduce acrylamide from coffee specifically, a darker roast is the lever, and it is a small one. See also our article on coffee and cancer risk and our article on mould and mycotoxins in coffee. Nothing here is medical advice.