The paint stripper in your decaf coffee

Almost nobody switches to decaf for fun. People switch because of sleep, anxiety, blood pressure, heart palpitations, a pregnancy, or a doctor's suggestion. It is a health decision. And almost nobody who makes it ever asks the obvious follow-up question: how did they get the caffeine out?

For a large share of the decaf sold in America, the answer is methylene chloride — the same solvent the EPA spent 2024 pulling out of consumer paint strippers because it causes cancer.

That is not a contamination scandal. That is the recipe, and it is legal.

The lawsuit that put this back in the news

On January 13, 2026, two consumers — Jennifer Williams of Ferndale, Washington and David Strauss of Irvington, New York — filed a proposed class action against Starbucks in the U.S. District Court for the Western District of Washington, case number 2:26-cv-00112. They are represented by Hagens Berman.

The complaint alleges that independent testing of Starbucks Decaf House Blend medium roast detected three volatile organic compounds. The numbers, as reported by the Spokesman-Review and Daily Coffee News:

Two things about that table matter, and most of the coverage skipped both.

The benzene comparison is to a water standard, not a food standard. The 5 ppb figure is EPA's Maximum Contaminant Level for benzene in drinking water. There is no equivalent federal limit for benzene in brewed coffee. EPA set the goal at zero because benzene causes leukemia, then set the enforceable number at 5 ppb because that is what treatment systems can realistically achieve. Comparing a coffee result to it is rhetorically effective and technically apples-to-oranges. We would rather say that plainly than let a number do work it cannot do.

The methylene chloride number is far below the legal limit — which is the actual story. Under 21 CFR 173.255, methylene chloride may legally remain in decaffeinated roasted coffee at up to 10 parts per million. That is 10,000 ppb. A result of 22 ppb is roughly 450 times under it. So the interesting question was never "did Starbucks break the rule." It is "why does the rule allow 10,000 ppb of this in the first place."

To be fair to Starbucks: the company said it firmly believes the allegations are inaccurate, the plaintiffs did not allege any violation of FDA residue limits, and the case has not been decided. Nothing here has been proven in court.

Methylene chloride is not an accident. It is a method.

There are four common ways to decaffeinate coffee, and the label is not required to tell you which one your bag went through.

• Direct solvent, also called the "European method." Steamed green beans are rinsed with methylene chloride, which binds the caffeine. The beans are then steamed again and roasted, which drives off nearly all of the solvent. This is the cheap, high-volume default.
• Ethyl acetate, sometimes marketed as "natural" or "sugarcane process" because ethyl acetate occurs naturally in fruit. It is still a solvent.
• Supercritical CO2, which uses pressurized carbon dioxide instead of an organic solvent. Expensive, mostly industrial scale.
• Swiss Water Process, which is water plus activated carbon filters and no solvent at all, taking beans to at least 99.9 percent caffeine-free.

Both National Geographic and Chemical & Engineering News have walked through the chemistry, and the honest summary is that roasting destroys the overwhelming majority of the residue. The disagreement is not really about whether 22 ppb will hurt you tomorrow. It is about whether a known animal carcinogen should be an approved processing aid in a beverage roughly one in ten American adults drinks every day, when a water-only method already exists and works.

The EPA banned it. The FDA still allows it in your cup.

In April 2024 the EPA finalized a rule banning most uses of methylene chloride and phasing out all consumer uses within a year. The Federal Register notice found the chemical presents an unreasonable risk of injury to health, including chronic cancer risk.

That rule was written under the Toxic Substances Control Act, which is EPA's authority. Food additives are the FDA's. So the same chemical is now off-limits in a can of paint stripper in your garage and still authorized in the coffee in your kitchen. The Center for Science in the Public Interest put it plainly: EPA banned methylene chloride, but it's still used in foods.

There is a petition sitting at the FDA. It has been there since 2023.

In late 2023 a coalition — the Environmental Defense Fund, Breast Cancer Prevention Partners, the Center for Environmental Health, the Environmental Working Group, and food-safety scientist Lisa Lefferts — petitioned the FDA to revoke food-additive status for four solvents: benzene, ethylene dichloride, methylene chloride, and trichloroethylene.

Their legal argument is the Delaney Clause, a 1958 provision barring the FDA from approving any food additive shown to induce cancer in humans or animals. Delaney is unusual in American food law because it has no "acceptable level" escape hatch. If the additive causes cancer, it is not supposed to be approved. STAT News covered the filing in 2024.

The FDA reopened the comment period specifically to ask manufacturers about the practical difficulty of phasing the solvents out. That comment period closed on June 29, 2026. As of this writing there is still no decision. The National Coffee Association is fighting it, arguing that about 10 percent of U.S. adults drink decaf daily and that banning European Method decaf would defy science and harm Americans' health.

Nearly three years. Four carcinogens. No ruling.

The cancer warning coffee used to carry, and then stopped

This is the part that tends to make people sit up.

In March 2018 a California court ruled that coffee sellers had to post Proposition 65 warnings about acrylamide, a compound formed when beans are roasted. Rather than let the warnings stand, California's Office of Environmental Health Hazard Assessment wrote coffee a specific exemption, which the Office of Administrative Law approved on June 3, 2019 and which took effect that October. The regulation states that exposures to chemicals created by roasting or brewing coffee do not pose a significant risk of cancer. Perkins Coie has a clear summary of how it happened.

Note carefully what changed and what did not. The acrylamide is still in the coffee. The epidemiology on coffee and cancer genuinely is reassuring, which is the defensible basis for the exemption. But the label went away by regulation, not because the chemistry changed.

What 7,069 tests on 45 coffee brands actually found

In August 2025 the Clean Label Project published the largest independent coffee screen we have seen: 7,069 tests across 57 products from 45 brands, sourced from Brazil, Colombia, Costa Rica, Ethiopia, Guatemala, Kenya, Peru and Hawaii. CNN reported it first.

Read that fairly. "Detected" is not "dangerous," the heavy metals came in under EU per-serving limits, and the study's own framing was that coffee is mostly safe. The findings worth changing behavior over are narrower and more specific:

• Every organic coffee tested still carried AMPA. Organic certification bars glyphosate use on the farm. It cannot bar spray drift, shared processing equipment, or contaminated water.
• Packaging shows up in the results. Phthalates ran highest in canned coffee and pods, lowest in bags. That is a straightforward argument for whole bean over pods.
• Medium roast came out worse than both light and dark. Not what most people would guess.

Clean Label Project lists Love Grown as the only coffee brand currently carrying its certification.

The one nobody prints on the bag

Ochratoxin A is a mycotoxin produced by molds that grow on coffee cherries during drying and storage. It is classified as a possible human carcinogen and is toxic to the kidneys.

How common is it? A 2019 global meta-analysis put the pooled prevalence in coffee and coffee-based products at 53 percent, with a pooled concentration around 3.2 µg/kg. A more recent 2023 systematic review of green coffee-based products estimated 38 percent, ranging from 0 percent in Thai samples up to 75 percent in Moroccan ones.

We want to flag something here. Higher figures than these get repeated online, including in our own short-form video on this topic. When we went back to the underlying meta-analyses, 38 to 53 percent is what they actually support, and that is the number we are standing behind. The EU enforces limits on ochratoxin A in coffee. The US does not have an equivalent enforceable limit, which is why a certificate of analysis from the roaster is worth more than the word "premium" on a bag.

How to tell what you are actually buying

The label will not tell you the decaffeination method unless the roaster chose to brag about it. So:

1. If the bag does not say Swiss Water, CO2, or ethyl acetate, assume it is solvent-decaffeinated. Roasters who pay for water process advertise it, because it costs more. Silence is information.
2. Buy whole bean, not pods or canned. That is the clearest signal in the Clean Label Project phthalate data.
3. Look for a certificate of analysis. A roaster publishing actual lab results for heavy metals, pesticides and mycotoxins is telling you something no marketing adjective can.
4. Do not assume organic solves it. For AMPA it demonstrably did not.
5. Keep the risk in proportion. The evidence on coffee overall is good, and switching to a water-process decaf costs a few extra dollars a bag. This is a cheap fix, not an emergency.

What we would actually want

Not a ban tomorrow. A line on the label.

Decaffeination method is a manufacturing fact the roaster already knows and the buyer currently cannot see. Every argument for keeping methylene chloride legal — the residue is tiny, roasting burns it off, the epidemiology is fine — is also an argument that disclosure should be easy. If the method is defensible, print it.

Until then the burden is on you to read a bag that was designed not to answer the question.

Tallow reads ingredient labels and surfaces published lab results so you do not have to hold four regulatory agencies in your head at the grocery store. Scan your coffee in Tallow and see which brands actually publish their numbers.

This post describes allegations in pending litigation, published lab screens, and current federal regulations. It is not a legal determination against any company, and it is not medical advice.