Is decaf actually caffeine-free?
No.
2–15 mg per cup
No. A cup of decaf carries roughly 2–15 mg of caffeine against about 95 mg in a regular cup. Here is what the measured figures actually say, and why the legal thresholds are looser than most people assume.
Sources on this page · McCusker et al. 2006 · FDA · Directive 1999/4/EC ·
No. Decaf is decaffeinated, not caffeine-free, and the difference matters if you are the kind of drinker who ends up on a site like this one.
The measured numbers
| Source | Decaf | Regular |
|---|---|---|
| FDA | 2–15 mg per 8 fl oz | 113–247 mg per 12 fl oz |
| National Coffee Association | 2–15 mg per 8 oz | 95 mg per 8 oz |
| McCusker et al. 2006, J Anal Toxicol | 0–13.9 mg per 16 oz brewed; 3.0–15.8 mg per espresso shot | — |
| Swiss Water | 3–15 mg brewed | — |
The safe summary: roughly 2–15 mg per cup, against about 95 mg for regular — somewhere between 1% and 15% of a normal coffee.
Two caveats do real work here. First, McCusker found that variability within a single outlet was large: six decaf espresso shots pulled at the same store ranged from 3.0 to 15.8 mg. Per-cup figures are not tight. Second, espresso is the most variable format, so if you are stacking evening decaf espressos rather than filter, you are working with the loosest end of the range.
The legal thresholds are looser than you think
This is the part most explainers get wrong, including several that rank well.
In the EU, Directive 1999/4/EC sets the decaffeinated threshold at “0,3% by weight of the coffee-based dry matter”. But read what that directive covers: it governs coffee extracts and chicory extracts — instant coffee. It does not set a caffeine limit for roasted beans.
The 0.1% figure everyone quotes for roasted decaf does not come from an EU-wide instrument that I could locate. It comes from national law and is applied de facto across the bloc; Germany’s stated threshold is 0.08%. The honest phrasing is: 0.3% for extracts under Directive 1999/4/EC, and roasted decaf governed at around 0.1% — 0.08% in Germany — under national rules rather than a single EU caffeine directive.
In the US, the widely repeated “97% removal is required by the FDA” appears to have no regulation behind it. The FDA’s own consumer page discusses decaf caffeine content and sets no threshold; the National Coffee Association frames 97% descriptively, as what the process typically achieves, not as a legal requirement. What US law does regulate is solvent residue: 21 CFR 173.255 caps methylene chloride at 10 ppm in decaffeinated roasted coffee.
The structural point is more interesting than either number. The EU regulates the outcome — how much caffeine is left. The US, in practice, talks about the process — what fraction was removed. Those are not interchangeable. A high-caffeine green coffee decaffeinated to “97% removed” can end up with more residual caffeine than a low-caffeine green put through the identical process.
What about “99.9% caffeine free”
That number belongs to Swiss Water. It is their guarantee for green coffee that has come out of their process — a supplier commitment, not a certification standard, and not a figure that transfers.
For comparison: sugarcane EA is documented by Cafe Imports at a minimum of 97% removal. Descamex say their Mountain Water process exceeds 97%, despite the 99.9% figure being widely attributed to them by roasters. CO₂ typically achieves 97–99%.
So a bag claiming 99.9% that is not Swiss Water is making a claim its decaffeinator does not.
If you are genuinely caffeine-sensitive
A few practical consequences:
- Filter over espresso, if you want the tighter end of the range. Espresso showed the widest per-shot variation in the only good study.
- Process does not predict residual caffeine in any useful way at these levels. All four commercial processes land in the same band.
- Naturally low-caffeine coffee is not decaf. Laurina sits around 0.6% caffeine by dry weight against roughly 1.2% for standard arabica — a third to a half, not a ninety-ninth. If you are avoiding caffeine for medical reasons, low-caf is not a substitute for decaf.
- Truly caffeine-free coffee does not exist commercially. Coffea charrieriana, described from Cameroon in 2008, makes theobromine instead of caffeine and is genuinely caffeine-free — and is a research plant with no product behind it.
If caffeine sensitivity is a medical question for you rather than a preference, the numbers above are population ranges from a small number of studies, not guidance — worth taking to a clinician rather than to a coffee site.
Sources
Questions
- How much caffeine is in a cup of decaf?
- The FDA gives 2–15 mg per 8 fl oz cup, against 113–247 mg in a 12 fl oz regular coffee. A peer-reviewed 2006 study measured 0–13.9 mg per 16 oz across ten outlets, and 3.0–15.8 mg per decaf espresso shot.
- Can decaf keep you awake?
- For most people, no — 2 to 15 mg is a small fraction of a regular cup. If you are unusually caffeine-sensitive, or drinking several decaf espressos late, the figures do stack, and espresso is the most variable format.
- What is the legal limit for calling coffee decaffeinated in the EU?
- EU Directive 1999/4/EC sets 0.3% anhydrous caffeine by dry weight, but it governs coffee *extracts* — instant coffee — not roasted beans. Roasted decaf is regulated at national level, commonly cited as 0.1%, with Germany at 0.08%. There is no single EU caffeine directive for roasted decaf.
- Does decaf have zero caffeine ever?
- Not from any commercial process. The only genuinely caffeine-free coffee species described, Coffea charrieriana from Cameroon, produces theobromine instead of caffeine — but it is a research plant, not a product.
Sources
Peer-reviewed · Journal of Analytical Toxicology 30(8):611, 2006
Does not say It does not establish a safe threshold for caffeine-sensitive people, or for pregnancy. It measured what was in the cup, nothing more.
Regulation · US Food and Drug Administration, 2024
Does not say The 400 mg figure is not a recommendation, a target, or a limit that applies to everyone. The guidance names pregnancy, medication interactions and individual sensitivity as reasons it may not apply.
Regulation · European Parliament and Council, 1999
Does not say It does not require any particular removal rate. A coffee that started low in caffeine and a coffee stripped by 99.9% can both sit under the same ceiling and carry the same word on the bag.