Decaf coffee contains caffeine. Not much, but not none. The FDA’s consumer page on caffeine carries the heading “‘Decaffeinated’ Does Not Mean Caffeine-free” and states: “Decaffeinated coffees and teas have less caffeine than their regular counterparts, but still contain some caffeine. For example, decaf coffee typically has 2 to 15 milligrams in an 8-fluid-ounce cup” (FDA, Spilling the Beans). For comparison, the same agency’s table puts regular brewed coffee at 113 to 247 milligrams per 12 fluid ounces. So decaf is a large reduction and not an elimination — and if you are drinking it because someone told you to avoid caffeine, that distinction is the point of this page.
Table of Contents
What this page used to say, and what was wrong with it
The number was too low. An earlier version of this page said, in three places, that “an 8-ounce cup of decaf has about 2 to 5 milligrams of caffeine”. The FDA’s range is 2 to 15. The bottom of the old range was right and the top was three times too optimistic, which matters most for exactly the people who care: someone drinking four cups of decaf could be taking in around 60 mg at the top of the real range, which is roughly a cup of black tea’s worth.
Two passages implied decaf has no caffeine at all. A “Health Benefits of Decaffeinated Coffee” table listed “Improved Sleep and Relaxation” with the explanation “Absence of caffeine aids better sleep patterns” — there is no absence. Another section described decaf drinkers as choosing “organic options for caffeine-free enjoyment”. Both are gone. The page also, correctly, said elsewhere that decaf “still has a bit of caffeine”, so it was arguing with itself; this version does not.
An unsourced safety claim about a solvent has been replaced with the regulation. The old page said: “Methylene chloride is safe in small amounts, but it may pose health and environmental risks.” That sentence asserts safety and hazard in the same breath, cites nothing for either, and an internal review of this site had already flagged it. It is dealt with properly below.
An unsourced health-benefits table has been deleted. It claimed decaf brings a “Lowered Risk of Heart Disease” and “Reduced Stomach Acidity”, with no study behind either, and it linked to a commercial health site rather than to research. So have three sentences that existed only to carry a link — “If you’re curious about other coffee-related health topics, you might wonder how long does coffee bloat last after drinking”, “Explore the impact of coffee on health to learn more about its effects”, and “If you’re curious about other coffee impacts, like drinking coffee after teeth whitening, more information is readily available”. None of them said anything. The FAQ block at the bottom has gone the same way.
Is there a legal definition of decaf?
In the United States, we could not find one for roasted coffee. The figure you will see everywhere is 97%, and it is worth being precise about where it comes from: it is the National Coffee Association’s description of what the process typically achieves, not a rule anyone enforces. The NCA’s wording is “The decaffeination process typically removes around 97% of the caffeine naturally found in coffee beans, leaving approximately 2–15 mg of caffeine in your cup” (NCA, Decaf coffee). “Typically removes” is a statement about industry practice. If you see it written as “the FDA requires 97% removal”, that is not what the NCA says, and no FDA rule to that effect could be found.
There is a US regulation in this area, but it caps a solvent residue rather than caffeine — see the next section. The practical consequence for a shopper is that “decaf” is a process description, and the caffeine that survives it varies by bean, by method and by how strongly you brew. The FDA’s 2-to-15-milligram range is the honest way to think about it.
Methylene chloride: what is actually known
The most common industrial route is what the trade body calls the European Method: green beans are steamed to open their “pores”, rinsed with “a liquid solvent, which bonds to and removes caffeine”, and then dried and roasted. The NCA’s page does not name the solvent — but the sentence in which it says the FDA “has set rigorous standards” for the European Method links to 21 CFR 173.255, the regulation governing methylene chloride, also called dichloromethane. That is the solvent in question. The NCA describes the solvent as being recovered and reused in “a highly efficient, closed system”, and says that across all processes the beans are afterwards “washed, steamed, and roasted at temperatures that evaporate any liquids used in the process.”
Here are the three facts we can put a document behind, and the reader can decide.
- There is a legal residue ceiling. 21 CFR 173.255(c) permits methylene chloride “In coffee as a residue from its use as a solvent in the extraction of caffeine from green coffee beans, at a level not to exceed 10 parts per million (0.001 percent) in decaffeinated roasted coffee and in decaffeinated soluble coffee extract (instant coffee).” That is the “rigorous standards” the NCA refers to. It is a limit on what may remain, not a verdict on what is safe at any dose.
- The substance itself is a listed carcinogen. The US National Toxicology Program’s Report on Carcinogens, 15th edition, lists dichloromethane as “reasonably anticipated to be a human carcinogen based on sufficient evidence of carcinogenicity from studies in experimental animals”. The same profile is equally clear about what is missing: “The data available from epidemiological studies are inadequate to evaluate the relationship between human cancer and exposure specifically to dichloromethane.” That listing is about the chemical, at occupational and experimental exposures. It is not a measurement of risk from drinking decaf, and nobody should read it as one. The NTP profile also notes that “due to health concerns, dichloromethane’s use as an extraction solvent in food products and coffee has declined greatly over the years” (NTP, Report on Carcinogens, dichloromethane profile, PDF).
- The rule is currently under petition. The FDA has before it a food additive petition from the Environmental Defense Fund and others asking it to remove four solvents, methylene chloride among them, from the food additive regulations. The agency reopened the comment period in May 2026 “to allow for the submission of any updated data and other information over the last two years”, with comments closing on 29 June 2026 (91 FR 31681, docket FDA-2023-F-5684, PDF). As of August 2026 we could find no decision. So the correct thing to say is that the authorisation stands and is being reconsidered.
One note on a phrase that circulates widely. The NCA states that the FDA “has said that risk to consumers from drinking European Method decaf is ‘essentially non-existent.'” We followed that citation: the NCA’s link for it points to a 2018 BBC article, not to an FDA document. The quote may well be accurate, but we could not verify it against an FDA source, so we are not repeating it as though we had. What we can point at is the regulation itself, quoted above.
Our position: if you would rather not have solvent-processed coffee, that preference is easy to act on and does not need to be justified — buy a bag that names Swiss Water, Mountain Water or CO2 on the label. If you drink it and are not worried, nothing above says you should be. What we are not going to do is what the old page did, which was to tell you it is “safe in small amounts” without showing you anything.
The four routes, and how to read a label
The NCA counts four methods, all of which start by wetting green beans and end with the caffeine somewhere else. Coffee is never labelled “methylene chloride decaf” — a bag with no process named on it is most likely solvent-processed, because the processes people pay extra for are the ones roasters advertise.
| What the label says | What it means |
|---|---|
| Nothing, or “naturally decaffeinated” | Usually a solvent process. “Natural” here refers to the solvent’s origin, not to an absence of solvent. |
| European Method, or a named solvent | Methylene chloride. Residue capped at 10 ppm by 21 CFR 173.255(c). |
| Ethyl acetate, sugarcane process, EA | A different solvent, one that occurs naturally in fruit and is produced from sugarcane fermentation for this use. |
| Swiss Water, Mountain Water | Water-based. No organic solvent is used. |
| CO2, supercritical CO2 | Carbon dioxide under pressure acts as the extractant. |
Swiss Water and other water processes
Swiss Water’s own description of what it does: “Our trusted process uses only water, temperature, and time to gently remove caffeine, while preserving all of a coffee’s origin characteristics”, and “we don’t use esters or solvents”. The beans meet a saturated green coffee extract, and “When the GCE surrounds the beans, diffusion allows only caffeine molecules to migrate outward, while all the precious flavor and other compounds remain exactly where they belong—in the coffee” (swisswater.com). That is the company describing its own process, which is the right source for how it works and the wrong source for whether it tastes better.
Two corrections here. The old page’s FAQ said the Swiss Water Process “uses water and osmosis to remove caffeine” — the company says diffusion, which is the different and correct mechanism: caffeine moves down its own concentration gradient into an extract that is already saturated with everything else. And the old page called water processes “chemical-free”, borrowing a marketing phrase; what is meant is that no organic solvent is involved. Mountain Water is a separately owned water-based process; the old page’s description of it was lifted from a supplier’s marketing and we have not verified it, so it is not repeated.
Ethyl acetate
Ethyl acetate is a solvent, and coffee decaffeinated with it is still solvent-processed however the bag words it. It occurs naturally in ripening fruit, and for coffee use it is commonly produced by fermenting sugarcane, which is why you see “sugarcane decaf” from Colombian producers. That origin is why the label often says “naturally decaffeinated” — a phrase describing where the solvent came from, not the fact that one was used. The old page called it “a more natural choice”; that is a description of provenance, not of safety, and it is not an argument that ethyl acetate is preferable to CO2 or water.
Supercritical CO2
Carbon dioxide held above its critical point behaves partly like a liquid and partly like a gas, and in that state it preferentially dissolves caffeine while leaving most of the larger flavour molecules behind. It needs high pressure and heavy equipment, which is why it shows up more in large commercial volumes than in small-lot coffee. One wording change: the old page said “The CO2 binds only with caffeine, removing it while keeping other flavors intact.” Selectivity here is a matter of degree, not an absolute, and “only” was doing work the chemistry does not support.
The efficiency table has been removed. It gave caffeine removal figures of 99.9%, 99%, 96–97% and 98% for four processes, plus a “Triglyceride Process” row, with no source for any of it. Those numbers are the kind of thing that gets copied between coffee pages until it looks like data. The only removal figure on this page now is the NCA’s “around 97%”, labelled as the trade body’s typical figure, and the only caffeine figure is the FDA’s 2–15 mg.
Does it taste different, and does it brew differently?
Decaf has a reputation for tasting flat, and part of that reputation is history rather than chemistry: for a long time the beans sent for decaffeination were not the good ones. Whether a specific bag tastes as good as its caffeinated equivalent is a question about that bag, and nobody at this site has cupped any of them, so there are no tasting notes here and the old page’s claims about brands “closely matching their caffeinated counterparts” have gone.
Two things are genuinely different and both are worth knowing:
- Decaf is harder to roast. The beans change colour differently, so roast level is harder to judge. Hottop’s roaster manual warns owners directly: “Extra care must be taken when roasting decaffeinated coffee. The decaffeination process changes the indicators used to judge roast level and over-roasting can result.” Some of the flat, burnt character people blame on decaffeination is over-roasting. If you roast your own, see roasting coffee at home.
- Brew it the same way as anything else. Filter brewing at 55 g of coffee per 1 kg of water — the standard brew ratio in SCA Standard 310-2021, §7.1.5, or 1:18 by weight. Stronger means more coffee, not less, and a coffee “cup” is 6 fl oz rather than 8. Water at 90–96 °C, with the sourcing on our best temperature for French press coffee page. There is no reason to under-dose decaf, and under-dosing is a common cause of the weak cup people then blame on the process.
One thing decaf does not change
If you switched to decaf for your cholesterol, the switch does not do that job. Cafestol and kahweol — the coffee oils that raise LDL cholesterol — survive decaffeination, and regular and decaffeinated coffees carry similar amounts of them. What controls how much reaches your cup is the filter, not the caffeine. Our coffee filter types page sets out the sources and the measured amounts rather than repeating them here. In short: a decaf French press and a caffeinated one deliver much the same diterpenes.
Who should treat the leftover caffeine as significant
Most people can ignore 2–15 mg. Some cannot, and they are the people most likely to be reading a decaf page: anyone advised to avoid caffeine in pregnancy, for a heart rhythm condition, for reflux, for anxiety or insomnia, or because of an interacting medication. The FDA’s own next sentence after the 2-to-15-milligram range is aimed at exactly this group: “If you react strongly to caffeine in a negative way, you may want to avoid these beverages altogether.”
For context on the other end of the scale, the FDA “has cited 400 milligrams a day — that’s about two to three 12-fluid-ounce cups of coffee — as an amount not generally associated with negative effects”, while adding that “there is wide variation in both how sensitive people are to the effects of caffeine and how fast they eliminate it from the body.” Our page on coffee beans for lattes corrects a product description on this site that once claimed a decaf coffee had “no caffeine”, which is where this correction started.
Disclaimer: This page is general information about coffee, not medical advice, and not a safety assessment of any chemical. The regulatory and toxicological documents quoted here are summarised, not interpreted for your situation. If you have been advised to limit or eliminate caffeine — in pregnancy, for a heart, blood-pressure, reflux, anxiety or sleep condition, or because of a medication — follow that advice and ask the clinician who gave it how decaf fits into it. Nothing on this page was tasted, brewed or measured by us.
