Green coffee: what it is and how it is judged

The seed, after the fruit and before the roast

Sourced

The FAO's guidelines on mould in coffee define a green coffee bean as the dried seed of the coffee plant, separated from the non-food tissues of the fruit, and note that coffee is exported in this form. The stage before it has its own name. Parchment coffee is wet-processed coffee after pulping and drying, still inside the hard, paper-like layer called parchment; the guidelines give its moisture as about 12 per cent. Coffee dried whole in its fruit goes by several names in the same document: cherry coffee, unwashed coffee or natural coffee.

The same document uses the word curing for the last stage of preparation, which usually happens just before sale for export: a series of operations that may include cleaning, polishing, screening, sorting and grading. So a lot of green coffee is not simply dried coffee with the husk off. It has been sorted, usually several ways, and what was sorted out has gone somewhere else.

From dried parchment to a shipped lot

Sourced

The steps are those listed in the International Trade Centre's handbook for the dry mill and after. It is a sequence, not a description of any one mill: the FAO's guidelines say it is impossible to generalise about who carries out each step, or when, because the trade is organised so differently from one country to the next.

What a buyer measures before tasting anything

Sourced

The handbook describes green assessment as a set of steps. The beans' colour and appearance are judged, with a check of their smell. Moisture and water activity are measured. Then come the two operations it calls paramount in any quality laboratory: a screen-size analysis and a count of defects. Elsewhere it lists the properties of green coffee that are taken to bear on the cup as density, moisture content, water activity, colour and smell, bean size, uniformity and the presence of defects.

Only after that is a sample roasted and tasted. The handbook says the main purpose of the sample roast is to uncover defects that the green inspection did not reveal, and it describes cupping as a practice at least a hundred years old that has been made as objective as training and calibration allow.

It is also candid about the limits. Coffee is rated by professionals following a protocol, it says, and the acceptable flavours are those preferred in western Europe, North America and Japan; these systems are, in its words, more rooted in the industry's habit and practice than in a clear body of research. That is the trade's own handbook describing its own methods, and it is the right frame for everything else on this page.

The moisture figures, and whose they are

Sourced

Moisture content is the share of a bean's weight that is water. Water activity describes how available that water is to moulds. Several bodies publish figures, for slightly different purposes.

These are limits set for safe storage and transport, not flavour targets, and the documents do not agree to the decimal. None is CoffeeHQ's figure. The trade handbook itself says 12.5 per cent is a technical point where the risk of damage becomes unacceptable, not a level every buyer will accept.

Why a moisture reading is less simple than it looks

Sourced

The FAO guidelines call moisture content the principal measure for predicting whether a lot will keep. They are careful about how it is usually taken. A moisture meter does not measure water; it estimates it from the electrical properties of the coffee, against a calibration made on a few samples whose moisture was found by drying them in an oven. Lots differ in ways that throw the estimate off, meters drift, and the document adds that they can be misused or deliberately mis-set. Its summary is that the measurement is deceptively simple and neither foolproof nor trivial.

It also says what the number cannot tell you: moisture content is a good predictor of storability but not of past handling. On that reasoning a lot that was wetted and dried again can read the same as one that never was.

Water activity: what it measures, and what it is not

Sourced

Moisture content is a quantity: how much of the bean is water. Water activity is a condition: how freely that water can leave the bean or be used by something living on it. It is written as a number between nought and one and is, in effect, the relative humidity of the air a sample would be in balance with, expressed as a fraction. A sample at 0.60 neither gains nor loses water in air at 60 per cent relative humidity at the same temperature.

That is why the measure exists. A mould does not respond to how much water a bean contains; it responds to whether it can take that water. The FAO guidelines accordingly state their thresholds in water activity, about 0.80 for production of the toxin and a little lower for any growth of the moulds that make it, and then translate them into moisture, the thing a mill can measure. The drying guide gives their figures for where the two meet.

The two measures are linked, but not by a fixed rule. The relationship between them, called a sorption isotherm, is a curve that shifts with temperature and differs between a coffee that is drying and one that is taking water back up. The guidelines say as much about their own figures: the relationship was measured only on coffee as it dried, and coffee that is rewetting might behave slightly differently. A Colombian study of 2025 measured the curve itself, though on parchment coffee and not on green: coffee still in the papery layer that is hulled off before export. It took eighteen samples of one arabica variety from one locality in Huila, sun-dried to between 9 and 11 per cent moisture, and measured them at 25, 35 and 45 degrees Celsius across water activities from 0.1 to 0.85. The curves are S-shaped. Over most of that range the coffee held less water the warmer it was, at the same water activity. Washed parchment held more water than parchment dried with its mucilage still on, which is a reminder that the curve belongs to the material measured. The authors publish their curves as charts and as fitted equations, not as a table of readings, so no figure is taken from them here, and a curve for parchment is not a table for a bag of hulled beans. Nor does the study settle the difference between drying and rewetting: its samples started near a water activity of 0.6, were dried below that and wetted above it, and the two halves were joined. A review of coffee seed physiology gives one point on the curve for living arabica seed: about 0.12 grams of water per gram of dry matter in balance with air at about 50 per cent relative humidity at 25 degrees Celsius. A Brazilian study that followed one natural coffee as it dried on a platform recorded the two measures side by side: a water activity of 0.82 at about 19 to 20 per cent moisture, 0.71 at 15.8, 0.63 at 12.9 and 0.52 at 11. Those are readings from one lot of fruit drying whole, on a moisture basis the paper does not state. They show the shape of the relationship and are not a table to apply to another coffee.

Three limits follow. A water-activity reading is meaningful only at the temperature at which it was taken. It describes the sample at that moment and, like moisture, says nothing about whether the lot was wetter last month. And a reading taken on a few beans from the top of a bag describes those beans.

It is not a flavour score. Water activity was devised as a measure of stability, and the documents read use it for exactly that: whether mould can grow. It is commonly expected in the trade that a lower reading means a coffee will keep its character longer. That is a reasonable expectation from the chemistry of storage, and it is an expectation: CoffeeHQ read no study that set lots arriving at different water activities side by side and tasted them months later. The 2023 storage study described in the guide to storing and shipping green coffee did measure water activity and taste the same coffee every three months, but on two coffees, which cannot show whether a lower reading on arrival means a longer life. A low number is evidence that a coffee is unlikely to go mouldy. It is not evidence about how it tastes.

Which measurements predict, which stand in, and which only sort

CoffeeHQ explanation

A measurement can do three different jobs. It can predict an outcome, with evidence that it does. It can stand in for something that cannot be measured directly, on an assumption. Or it can sort lots into classes that buyers have agreed on, whether or not the classes mean anything in the cup. The sorting below is CoffeeHQ's, from the documents read for these pages.

Moisture content

Predicts. The FAO guidelines call it the principal measure for predicting whether a lot will keep, and add that it predicts storability and not past handling.

Water activity

Predicts mould risk, on the guidelines' thresholds. Stands in for flavour stability when a contract uses it that way, on an expectation nobody read here has tested.

Defect count

Sorts, by a weighted convention that differs between national systems. It predicts the cup only for the defects that taste, and the guide to grading sets out how little has been tested.

Screen size

Sorts. The trade handbook's reason for sizing is that beans of one size roast evenly. No document read shows that a larger bean tastes better.

Density

Stands in. The trade handbook describes dense beans as the best, and one Ethiopian study found density rising with elevation, but that study tasted nothing. Density is used as a sign of slow growth and full development, which is an inference.

Colour and smell of the green bean

Stand in for freshness and sound processing, on a trained judge's experience. The handbook lists both among the properties taken to bear on the cup and describes the whole system as rooted more in habit than in research.

A test for 16-O-methylcafestol

Predicts, in a narrow sense: this compound is found in robusta and not in arabica and survives roasting, so its presence shows robusta in a coffee sold as pure arabica. It answers one question and no other.

Free fatty acids

A laboratory index of chemical ageing that rises with storage temperature and moisture. It stands in for staleness; the experiment that measured it reported no tasting scores, only that coffee kept at 40 degrees Celsius for a year made brews its authors called not worth discussing.

The sample roast and cupping

The only assessment that measures the thing a buyer cares about. It predicts how that sample tastes on that day, to those tasters, and depends on how well the sample stands for the lot.

Three ways a sample stands in for a shipment

Sourced

Coffee is usually sold before the buyer has seen it. The trade handbook describes three arrangements.

Subject to approval of sample

The exporter must send a sample before shipment, and the buyer is not bound to take coffee they have not approved. The handbook gives three forms: no approval, no sale; a repeat basis, where a second or third sample may follow a rejection; and several samples sent at once for the buyer to choose from.

Stock-lot sample

The sample is drawn from a parcel already in store, so shipment and sample should not differ. The handbook calls this usually the best route for a new exporter or for premium coffee, and too cumbersome for everyday trade. A later sale may be offered as quality equal to that stock lot.

Type sample

After several satisfactory deals the two sides agree that a quality is a type, and trade it without samples. The type is usually kept confidential between them.

Lot, micro-lot, crop year: words without a rulebook

Sourced

A lot is a quantity of coffee kept and sold as one thing. The FAO guidelines treat it as the unit that is sampled, and warn that a lot may itself be a blend of sources, which is why they ask for every bag to be speared when a sample is drawn. The handbook uses micro-lot throughout for small, separately sold parcels and never defines it. It says of its own three market segments, standard, premium and specialized, that the terms have no globally recognised and quantifiable definitions, and the same is true of the word micro-lot in every document read for this page.

What the handbook does say is what separation costs. A micro-lot is more labour- and resource-intensive to produce than a bulk blend, and it takes more sampling and selling effort for each kilogram sold. In its example from Ethiopia, contracts for micro-lots may cover only five to ten per cent of what a farmer grows; the rest is likely to be sold in the lower-priced segments.

Traceability is described in its tables as a matter of degree: little or none for standard coffee, where the concern is food safety, and tracing to the farm or plot for the most specialised. It is a record of where coffee has been. It is not in itself a statement about flavour.

Dates are looser still. The handbook notes that 1 October is the first day of the coffee year the International Coffee Organization uses for its statistics, and says coffee is usually best marketed when it is still fresh, which it glosses as new crop. No document read for this page defines past crop or old crop, and none says how many months separate one from the other.

What you can tell from a bag on your own table

CoffeeHQ explanation

A home roaster has none of the instruments above and can still do the first step of an assessment, which is to look and to smell. Evenness of size and colour matters for the reason the mills give: mixed beans roast unevenly. Black beans, insect holes, fragments and anything that is not coffee are visible without training. A musty or sack-like smell is a warning that needs no meter.

What cannot be done is what the trade cannot do either: tell from the green bean how the coffee will taste. That takes a small roast and a cup, which is the reason professional buying is built on samples and not on descriptions.

What this page leaves out

CoffeeHQ explanation

It gives no moisture, water-activity or density figure as a target. Each figure above belongs to the body named beside it and was set for storage and safety.

It does not define past crop, old crop or micro-lot, because no document read defines them. It does not describe how density is measured or what figures are considered high, because the handbook names density as a criterion without giving a method. It says nothing about what to buy, from whom, or at what price.

The international standards for sampling, moisture and size analysis were not opened; the publisher's pages refused the request. They are named here only where another document names them.

What to do next

• Coffee grading, system by system — What each country's grade actually counts, and why two grades with the same name can mean different things.
• Storing and shipping green coffee — How green coffee gains and loses water in a warehouse and in a container, and what studies of packaging found.
• Drying coffee — Green coffee starts where drying ends; this covers how coffee gets to that point and the limits it is dried to.
• Roasting coffee at home — The next step for a reader holding a bag of green coffee, with the makers' safety instructions.