Ochratoxin A is a compound made by a handful of moulds that can grow on coffee when the crop dries too slowly, catches rain part-dried, or sits too damp afterwards. A number of jurisdictions regulate ochratoxin A as a contaminant in roasted and soluble coffee, testing is routine in the trade, and the controls that matter are agricultural: they happen in the drying yard, the warehouse and the shipping container, not at a kitchen counter.
That last point is where the question usually goes astray. People ask which coffee has ochratoxin in it, as though it were a property of a bean, an origin or a roast style. It is not. It is a property of how one particular lot was treated during the weeks and months when it was vulnerable.
What ochratoxin A actually is
Ochratoxin A is a mycotoxin: a secondary metabolite, a compound certain moulds make as they grow. It is not part of the coffee plant, and it is not created by roasting or brewing. The moulds that produce it are ordinary storage fungi found on dried agricultural goods everywhere, turning up on cereals, dried vine fruit, cocoa, nuts and spices as readily as on coffee. Nothing about coffee attracts them. Coffee simply spends a very long time in the conditions they like.
In coffee the producers come from two groups within Aspergillus. One is the ochraceus group of yellow-green species, which includes Aspergillus ochraceus and close relatives such as A. westerdijkiae and A. steynii. The other is the black aspergilli, chiefly A. carbonarius, with A. niger described as an occasional and much less consistent producer. One thing is worth knowing when reading older material: the ochraceus group was reorganised taxonomically after a good deal of the early coffee work was published, and isolates once recorded as A. ochraceus have since been reassigned to species described later, which are reported to be the stronger producers of the toxin. Sources therefore differ on which organism dominates in which growing region, and part of that disagreement is bookkeeping rather than biology.
Why this is a handling problem, not a coffee problem
Moulds are not choosy about the crop. What governs whether they grow, and whether a growing colony makes toxin at all, is the water available to them, the temperature, and how much time they get. Give a mould free water, warmth and time and it will colonise almost any dried commodity; take the water away and it stops, whatever the commodity is.
The variable that matters is water activity, meaning how available that water is to a microbe, rather than moisture percentage. The two are different measurements and moulds respond to the first; water activity is covered in coffee water activity, and moisture content, which governs trading specifications and roast behaviour, in coffee moisture content. A lot can sit comfortably inside a moisture specification and still hold a water activity that storage fungi find workable, which is why a percentage on its own is a weak guide to whether a lot is biologically stable.
Now set coffee's timeline against that. A cherry is picked wet. It then spends days or weeks losing water on a patio, a raised bed or in a mechanical drier, outdoors, in humid tropical air, exposed to the weather and to overnight dew. Once dry it goes into bags, and those bags may sit in a warehouse at origin, then in a container crossing an ocean, then in an importer's warehouse, then in a roastery, for months and sometimes years before anyone drinks it. That is an unusually long stretch of vulnerability for a dried food, and it is precisely the window in which a slow-drying, interrupted or re-wetted lot becomes hospitable.
Two lots from the same farm can end up on opposite sides of that line. Take the first: picked ripe, pulped or spread the same afternoon, laid thin on raised beds, turned through the day, covered at dusk against dew and at any hint of rain, brought down to a stable target without stalling, rested in a ventilated store, bagged onto pallets away from the walls, and loaded into a lined container. Nothing in that sequence offers a storage mould a useful window. Now the second: picked with a proportion of overripes and ground-fall, held overnight in a heap because the mill was full, spread thick because there was no room on the beds, stalled through several cloudy humid days, rained on once while still leathery, bagged a shade early to clear the yard, stacked against an outside wall, and shipped without much thought about condensation. Same farm, same variety, same altitude, same year, and only one of those two lots ever offered anything to grow on.
The practical consequence follows directly. The entire control strategy is to get the crop dry evenly and reasonably quickly, and then to keep it dry. That is the same discipline that protects flavour. Coffee dried too slowly picks up fermented and phenolic notes; coffee dried unevenly ages unevenly; coffee stored damp or warm fades, goes woody and loses acidity long before anything dramatic happens to it. A supply chain that keeps water under control gets both outcomes for a single effort, which is why the useful answer here is structural, unglamorous and entirely upstream of the person drinking the coffee.
It also explains why origin, variety and roast style are such poor predictors. None of them changes the water, the temperature or the time. A meticulously dried natural from a hot, humid lowland can be cleaner than a carelessly handled washed lot from a cool highland, because the fungi respond to the handling rather than to the label. Be careful, then, with the idea that origin or roast level sorts coffees into categories on this point; what actually separates two lots is how well the water was managed between the tree and the roastery.
Where in the chain it can form
Risk concentrates at a short list of points, and every one of them is about water and time.
- Cherries that reach the ground. Overripe fruit that drops, or cherry swept up off bare earth, arrives already carrying soil fungi and often already split.
- Slow or interrupted drying. Layers piled too deep, too little turning, drying that stalls in humid weather, or lots left uncovered overnight so they take moisture back on.
- Rain landing on drying coffee. A shower on part-dried cherry is the classic trigger: the surface rewets while the interior is still moist, and in dry (natural) processing the outer fruit layer is exactly where colonies establish.
- Damaged and insect-bored beans. Broken beans, bored beans and husk fragments give fungi a way past an intact seed coat. Work on green coffee has repeatedly pointed to the broken, infested and husk fractions as the important carriers.
- Re-wetting after drying. Bagging before a lot has finished, stacking bags on a damp floor or against an outside wall, and condensation inside a container that travels from warm humid air into cool air: each one puts liquid water back onto a dry seed.
Drying method matters mostly through how much control it gives over that timeline; patio, raised-bed and mechanical drying each behave differently when the weather turns, and the trade-offs are set out in coffee drying methods. Storage matters through ventilation, stacking, bag choice and protection from condensation, which is the subject of green coffee storage. Surveys have generally reported higher occurrence in dry-processed than in wet-processed coffee, which fits the mechanism, since cherry dried whole keeps a sugary, slow-drying fruit layer wrapped around the seed and takes longer to come down. Washed coffee is not exempt, though, and the processing label on its own predicts far less than the handling behind it.
Why it hides in a small fraction of the lot
Mould-derived contamination in coffee is not spread evenly through a bag. It is concentrated, sometimes very sharply, in a small number of beans: the ones that were damaged, that sat in a wet pocket of the drying bed, or that came off the ground. Everything else in the same lot may carry none at all. That lumpiness has three practical consequences, and together they explain most of what the trade does about this.
First, it makes sorting genuinely worthwhile. Density separation, screen cleaning and optical colour sorting pull out blacks, sours, broken beans and husk, which are the fractions most likely to be carrying, and removing them lowers the average for the lot. Hand sorting at origin does the same work more slowly. This is one of the few places where an ordinary quality operation and a contaminant control are the same operation.
Second, sorting is a reduction rather than a removal. A bean can look entirely normal and still carry the compound, because the mould that made it may have grown and then died back without leaving an obvious mark; and plenty of visually mouldy beans carry none at all. Visual and sensory grading is a proxy, not a test, and treating a clean-looking screen as proof is a mistake.
Third, lumpiness makes sampling hard. If a handful of beans in a container carries most of the load, a small sample can easily miss them altogether or hit them and overstate the lot. That is why sampling protocols for this contaminant are so prescriptive about drawing many increments right across a consignment, combining them, and grinding the composite finely before analysis. It is the arithmetic of an unevenly distributed contaminant rather than bureaucratic fussiness, and it is also why two analyses of the same shipment can disagree if the sampling was done casually.
What a taster actually perceives
Nothing, in the specific sense. The compound is reported to have no flavour or aroma of its own at the levels found in commercial coffee, and cupping is not a way of detecting it. What a cupper can detect is the handling that often accompanies it: mouldy and musty notes, damp cellar and wet cardboard, earthy or dirty flavours, phenolic and medicinal edges, and the sharp fermented character of a stinker bean. Those are real cup faults with their own causes, and they are worth removing on their own terms.
They are also a loose signal rather than a dependable one. Lots that cup mouldy are more likely to carry more; lots that cup clean can still carry some. That is exactly why the trade tests for this rather than tasting for it. The distinction is worth stating plainly: this is a contaminant and regulatory question rather than a flavour fault, and it is handled with laboratory analysis and supply-chain records rather than with a spoon.
At a glance: where it can and cannot form
| Stage | State of the coffee | Can it form here? | What controls it |
|---|---|---|---|
| Ripe cherry on the tree | Wet, intact fruit on a living plant | Rarely; damaged, overripe or dropped fruit is the exception | Picking ripe, picking on time, keeping fruit off bare soil |
| Harvest and delivery to the mill | Wet fruit, warm, often in heaps | Yes, if fruit sits in heaps before processing | Moving fruit to the mill or the drying bed the same day |
| Drying (patio, raised bed, mechanical) | Falling from wet to dry over days or weeks | Yes, and this is the main window | Thin turned layers, cover against rain and dew, drying to target without stalling |
| Green storage and shipping | Dry seed, months to years | Yes, if the lot is re-wetted or held humid | Ventilated stores, pallets, sound bags, condensation control in containers |
| Cleaning, grading, sorting | Dry seed | No formation; removal of carriers | Density, screen and colour sorting; taking out husk and broken beans |
| Roasting | High heat, minutes | No formation; partial breakdown | Reported to reduce but not to eliminate, and published reductions vary widely |
| Packed roasted coffee | Dry, low water activity | No, unless the pack is wetted | Keeping roasted coffee dry and sealed |
| Brewed coffee | Water, minutes | No | Drunk immediately; splits between the drink and the spent grounds |
Roasting, packing and the cup
Roasting breaks some of it down. How much is genuinely unsettled: published reductions run from marginal to nearly complete, depending on roast level, roasting method, how much was present to begin with, and how the measurement was made. Some is also carried away physically with the chaff. Because the reported range is so wide, no roast profile is treated as a removal step by anyone in the chain, and a precise figure for how much a roast destroys claims more than the published record supports.
Roasted coffee in a sealed bag is a dry, low-water-activity environment in which storage moulds do not grow. A roasted bag that gets thoroughly wet and stays wet can grow visible mould, but that is ordinary spoilage you can see, and a different problem with a different answer. Brewed coffee is water, but it is drunk within minutes; what varies is how much of the compound moves into the liquid and how much stays behind in the spent grounds, and that split is reported to differ by brewing method, with contact time and the volume of water passing through the bed doing most of the work. Soluble coffee behaves differently again, because it is a concentrated extract rather than a ground seed, which is why jurisdictions that set maximum levels generally set a separate one for it.
The regulatory picture, and what mould-free claims mean
A number of jurisdictions set maximum levels for ochratoxin A in roasted and in soluble coffee, with the figure differing between those two product forms and between jurisdictions. Others set no numeric limit at all and rely instead on general powers over contaminated food. Green coffee is frequently not covered by a published threshold in the same way, which surprises people: control at that stage lives in codes of practice and in commercial contracts instead. There is an internationally agreed code of practice for the prevention and reduction of this contaminant in coffee, and from beginning to end it is a document about picking, drying, storage and transport rather than about the cup. Testing is routine, in the sense that exporters, importers and larger roasters run analyses on lots and buyer specifications commonly require it. Published survey figures vary a great deal between regions, product forms and years, so any single number is best read as one snapshot rather than a general fact.
Coffee is sometimes marketed as mould-free, mycotoxin-tested or clean. Such claims exist, and they range from a published laboratory report to a phrase printed on a bag. The underlying controls are not proprietary and never have been: pick ripe, dry properly, store dry, sort, test. They are ordinary agricultural and warehousing practice, available to any functioning supply chain, and used routinely by a great many roasters who never mention them on the packaging. Where such a claim rests on laboratory analysis, the analysis is the same routine testing already used across the trade.
What each part of the chain realistically does
- Producers and mills: pick ripe rather than stripping, keep fruit off bare ground, process or spread the same day, dry in thin turned layers, cover against rain and night dew, dry to target without stalling, and rest the coffee somewhere ventilated and dry.
- Exporters and shippers: keep stores and bags in good condition, keep coffee off floors and away from outside walls, protect containers against condensation, and decline to load lots that have not finished drying.
- Importers and roasters: specify moisture and water activity on arrival, draw proper multi-increment samples, test where the contract requires it, sort out or reject the fractions that carry risk, and hold green stock in stable, dry conditions.
- Drinkers: realistically nothing beyond buying from a chain that does the above and keeping coffee dry and sealed at home. Testing is a laboratory job rather than a kitchen one, no brewing method removes the compound, and no roast level settles the question.
The bottom line
Ochratoxin A is real, it is monitored, and it is a handling problem. It forms when moulds get water, warmth and time on a crop that is drying too slowly or sitting too damp, which makes it a question about drying yards, warehouses and containers rather than about beans, origins or roast styles. Sorting reduces it, roasting reduces it unpredictably, and neither is a control anyone leans on; keeping the coffee dry is. That is the same discipline that keeps coffee tasting like itself, so nobody has to choose between the two.