Foundational
Decaf coffee, explained
How caffeine is taken out of green coffee, what regulators say about the solvents, and how much caffeine a decaf still contains.
In short
Decaffeinated coffee is ordinary coffee that had most of its caffeine taken out while it was still green, before roasting. The caffeine is dissolved out of the raw bean with one of three kinds of solvent: an organic liquid, water, or carbon dioxide under pressure. Every process leaves a little caffeine behind, and every process takes some other things with it.
This page sets out what each process is, what two regulators say about the solvents, and what was measured in the cup. It does not rank the processes, because no comparison of them made on the same coffee was read.
It happens to the green bean
Sourced
Decaffeination is carried out on green coffee, before roasting. A 2026 study of decaffeinated coffee describes it as a technologically demanding step and lists the technologies in use: extraction with organic solvents, a water process, extraction with carbon dioxide above its critical point, and newer experimental methods. An earlier paper groups them the same way, as three approaches, using organic solvents, water, or liquid or supercritical carbon dioxide, and names the two solvents commonly used: methylene chloride, also called dichloromethane, and ethyl acetate.
The same papers make a point that the name hides. The processes are designed to take out caffeine selectively, and they also affect other constituents of the bean. What changes, and how much, is taken up further down this page.
Three routes to the same bean
Sourced
A diagram of the families of process, not a description of any one factory. The detail under each comes from the document named in it.
Green coffee, wetted
All three routes work on raw beans that have been moistened or steamed. One patent for the carbon dioxide route specifies raw coffee and moist gas; one company's account of its water process begins with cleaning and hydrating the beans.
Route one: an organic solvent
Methylene chloride or ethyl acetate dissolves the caffeine. Both are named in United States food regulations as permitted for decaffeinating coffee, on the conditions set out below.
Route two: water
One company describes immersing the beans in a caffeine-free extract of green coffee, so that caffeine moves out of the bean, then passing that extract through carbon filters that trap the caffeine so it can be used again.
Route three: carbon dioxide
A 1973 patent describes passing moist carbon dioxide, held above its critical temperature and pressure, through raw coffee, and then removing the caffeine from the gas with active carbon or by letting the gas expand.
Drying, then roasting
The decaffeinated green coffee is dried back and sold on to be roasted like any other. Everything a roaster and a brewer do afterwards is the same in kind.
A model of the three families of decaffeination. The papers read name these families; the steps are from one patent and one company's description, and real plants differ in ways no document read here describes.
What the carbon dioxide patent actually specifies
Sourced
The patent, filed in the United States in 1973 by Kurt Zosel on a German application of 1970 and granted in 1981, is worth reading for its numbers because the method is often described loosely. It gives as preferred conditions a temperature of about 40 to 80 degrees Celsius and a pressure of 120 to 180 atmospheres, with decaffeination taking from 5 up to about 30 hours, and its worked example runs at about 70 degrees and 160 atmospheres. The caffeine is held on active carbon while the carbon dioxide and water pass through; the patent adds that it can instead be separated by letting the gas expand.
A patent describes the process its inventor claimed, not what an industry does. These conditions show the scale of the thing, a pressure vessel run for many hours, and they are not a specification for how decaffeinated coffee is made today.
What two regulators say about the solvents
Sourced
Each row is one regulation's wording for one solvent. They are limits on what may remain in the coffee, set by different authorities in different units.
| Variable | Typical range | What moving it does |
|---|---|---|
| Methylene chloride, United States | not to exceed 10 parts per million | In decaffeinated roasted coffee and in decaffeinated soluble coffee extract (instant coffee). Title 21 of the Code of Federal Regulations, section 173.255(c), as the electronic code stood on 7 October 2026. |
| Dichloromethane, European Union | 2 mg/kg in the roasted coffee | Directive 2009/32/EC, Annex I, Part II, in the text consolidated on 16 February 2023. The same substance as methylene chloride. |
| Ethyl acetate, United States | no figure; good manufacturing practice | Section 173.228 permits it as a solvent in the decaffeination of coffee and tea and sets no numerical residue limit. |
| Ethyl acetate and carbon dioxide, European Union | no figure; good manufacturing practice | Annex I, Part I lists both among the solvents that may be used for all purposes in line with good manufacturing practice. |
| Methyl acetate, European Union | 20 mg/kg in the coffee | Annex I, Part II, for decaffeination of coffee and tea. Ethylmethylketone carries the same figure. |
These are regulatory limits on residues, read in the regulations themselves on 9 October 2026: the United States sections on the government's electronic code, current to 7 October 2026, and the European annex in the consolidated text of 16 February 2023, the latest CoffeeHQ could open. The three amendments made to the directive since 2009 did not touch the coffee rows; rules change, so check the current text before relying on a figure. They are not measurements of any coffee, they say nothing about the amounts actually found, and this page draws no conclusion from them about health. A part per million and a milligram per kilogram are the same proportion, so the two limits on methylene chloride differ fivefold.
What 'decaffeinated' has to mean, where a rule was found
Sourced
Two definitions were read. The European one covers coffee extracts, meaning instant coffee and liquid or paste concentrates: under Directive 1999/4/EC the label carries the word decaffeinated on condition that the caffeine does not exceed 0.3 per cent by weight of the coffee-based dry matter. That directive does not govern roasted coffee or brewed drinks.
For roasted coffee the one rule found is national. Germany's coffee ordinance reserves the word for green and roasted coffee containing at most one gram of caffeine in a kilogram of coffee dry matter, which is 0.1 per cent, and for extracts at most three grams in a kilogram. It is one country's rule, read in German in the ordinance's own text; CoffeeHQ did not look for the equivalent in every country and no European Union rule for roasted coffee was found.
A figure of 97 per cent of caffeine removed is often quoted as the requirement for decaffeinated coffee, usually as the United States rule. CoffeeHQ did not find it in any regulation it opened, which is not the same as saying none states it, and it is not repeated here as a rule. The two United States sections above set conditions on the solvents and do not define the word.
How much caffeine a decaf still has
Sourced
Decaffeinated coffee is not free of caffeine. Surveys of decaffeinated drinks bought in shops have found anything from none detectable to a little under 18 milligrams in a serving. Those surveys were read from their abstracts only, which is why CoffeeHQ's caffeine estimator gives decaf as a range of 0 to 18 milligrams per serving and marks it as inferred.
One measurement was read in full. The 2026 study described below brewed strong filter coffee, 10 grams to 100 millilitres, from an arabica and a robusta and from the same suppliers' dichloromethane-decaffeinated versions. The ordinary brews held about 760 and 1,110 milligrams per litre; the decaffeinated ones about 50 and 30. That is one pair of coffees and one process.
The estimator's range is an estimate from a few surveys, not a property of any cup. A larger serving, a stronger brew or a different product can sit anywhere in it. Anyone who has been told by a clinician to avoid caffeine altogether should treat decaf as containing some.
What comes out with the caffeine
Sourced
Three studies were read on this, and each measured something different. The 2026 study compared brews of an arabica and a robusta with their dichloromethane-decaffeinated counterparts, all roasted to the same colour. The change did not run one way. The decaffeinated brews held more chlorogenic acids and more trigonelline than the ordinary ones, and less of the oily fraction: diterpenes, tocopherols and fatty acids. The authors suggest decaffeinated beans reach a given colour in a shorter roast and so lose less in the roaster, and they note that another study found the opposite for chlorogenic acids.
A 2022 study analysed the air above eight pairs of regular and decaffeinated coffees sold in capsules by five vendors. In the three pairs matched for origin, species and roast, alcohols, pyrazines, phenols, lactones, pyrroles, pyridines and sulphur-containing compounds gave larger signals in the regular coffees, and furans, acids, ketones and aldehydes gave larger signals in the decaffeinated ones. The vendors had used different processes, among them two carbon dioxide routes; the study separated regular from decaffeinated and did not set out to compare one process with another. Nobody tasted the coffees, and the authors ask for more samples before firm conclusions are drawn.
A third study put regular and decaffeinated espresso, one commercial product sold in both versions, before thirteen trained assessors. Every sample was sweetened, with sugar, sucralose or stevia, because sweeteners were what the study was about. The regular espressos scored higher for roasted aroma, bitterness and the creaminess of the crema; the decaffeinated ones for acid taste, and they measured slightly lower in pH. The process used on the decaffeinated version is not stated, so the study describes that product and cannot say what any one process does.
None of these compares one process with another on the same beans. The claim that one route keeps flavour better than another is made often, by sellers of each, and nothing read here tests it.
Words on a bag of decaf
Accounts differ
- Water process, and the branded versions of it
- A description of the solvent used, which is water. Brand names belong to particular companies and describe their own process; one such company calls its method chemical-free, which is its wording and not a regulatory term.
- Sugar cane process, or natural decaffeination
- Sellers use these names for decaffeination with ethyl acetate, on the ground that the solvent occurs in fruit and can be made from fermented sugar cane; no document read confirms where any seller's solvent comes from. It is a solvent process under both sets of regulations above, and no document read defines the word natural for decaffeination.
- Carbon dioxide or sparkling water process
- The pressurised carbon dioxide route. Sellers' names for it vary and no standard wording was found.
- No process named
- Many bags do not say. Neither the United States sections nor the European annex read here requires the process to be declared on the label.
If a decaf brews differently from what you are used to
CoffeeHQ explanation
These are practical suggestions. Baristas commonly report that decaffeinated coffee behaves differently in a grinder and a brewer; no measurement of that was read.
- Espresso runs faster or slower than your usual coffee at the same setting
Usually because: Any change of coffee changes how a grinder setting behaves, and a decaf is a different coffee.
Change first: Dial it in as a new coffee, moving the grind one step at a time.
- The cup tastes flat compared with the same roaster's regular coffee
Usually because: It may be the coffee and not your brewing: one study above found smaller signals from several families of aroma compounds in decaffeinated coffees, and larger signals from others.
Change first: Try a different decaf before changing your recipe.
What nobody here ranks or promises about decaf
CoffeeHQ explanation
It does not say which process tastes best, which is safest or which is most natural; no document read supports any of the three. It gives no percentage of caffeine removed. It makes no statement about decaffeinated coffee and health, and gives no advice on how much anyone should drink. It does not describe how present-day plants are run, since the two process descriptions are a patent from 1973 and one company's own account. And it does not say that decaffeinated coffee goes stale faster or grinds differently, which are widely repeated and were not found in anything opened. On roasting, the only thing read is the one study's suggestion, made in passing, that decaffeinated beans reach a given colour sooner.
What to do next
- Caffeine in coffee — What sets the caffeine in an ordinary cup, and how widely measured servings vary.
- Caffeine estimator — Estimates a range for your own recipe, with decaf and half-caf as options.
- How the caffeine estimates are made — The measured studies behind the ranges, and where they are thin.
- Instant coffee — The one product for which a legal definition of decaffeinated was found.
- Coffee chemistry — What chlorogenic acids, trigonelline and the aroma compounds are.