What 8,687 Coffee Studies Have Actually Tested in People

Every claim made for coffee comes with the word "studies" attached. Studies show it protects the liver. Studies show the mushrooms in it sharpen focus. The word does a great deal of work and is almost never unpacked, because unpacking it requires knowing how many studies, in what, and of what kind.

So we counted. Our research index holds 8,687 coffee and caffeine papers published between 1990 and 2026, pulled from OpenAlex^[1], each with an abstract. We sorted every one by whether it reports work in people or work in animals and cells, then grouped them by the claim they address.

The picture is far more uneven than the marketing suggests.

How much of each claim was tested in people

Exercise performance is the best-tested claim in coffee science. Of 396 papers, 246 report human participants and 189 are randomised or controlled. That is what a well-tested claim looks like: most of the work done in the population the claim is about.

Cancer is the most-written-about and among the least tested. It has 994 papers, more than any other claim here, but only 24 randomised or controlled ones, against 344 in animals or cells. The volume of literature and the strength of the evidence point in opposite directions.

The gut microbiome claim is almost entirely pre-clinical. Forty-six papers, two reporting human participants. It is one of the most confidently stated coffee claims online and one of the least examined in people.

The mushrooms are a different order of thin

Functional mushroom coffee is sold on research. Here is the research.

Lion's mane is the most-marketed of them and has six papers in our index reporting human participants, of which five are randomised or controlled. The best known is a 2009 double-blind trial in 30 older Japanese adults with mild cognitive impairment^[2]. It is a real trial and a genuinely interesting result. It is also thirty people, in one country, in one population, fifteen years ago.

Chaga has three. Cordyceps five. Turkey tail two. Reishi, the best studied, has 24 out of 356.

Caffeine, in the same cup, has 1,289 papers reporting human participants and 646 randomised or controlled ones.

None of that means the mushrooms do nothing. It means that if you are buying mushroom coffee because it is "backed by science", the science backing it is overwhelmingly in mice and cell cultures, while the ingredient with the enormous human literature behind it is the coffee.

Why the distinction matters

A compound that does something in a cell culture often does nothing measurable in a person at a realistic dose. Absorption, metabolism and the sheer difference between a concentrated extract applied to cells and a scoop stirred into a mug all intervene. Pre-clinical work is how researchers establish that an effect is possible and by what route. Human trials are how anyone establishes that it happens in practice.

Both are real science. Only one supports a sentence on a package.

The literature is growing and thinning at once

Coffee research has expanded roughly twentyfold since the early 1990s. Average citations per paper have fallen by about half over the same span, which is what you would expect as a field grows: more work, spread thinner, with attention concentrating on a smaller share of it.

What this does not tell you

Three limits, stated plainly because the whole point of the exercise is to be checkable.

We classified from abstracts, not full texts. A paper that mentions mice once is counted as animal work. A paper whose abstract omits its design is counted in neither column.

Claims were matched by abstract text, which over-counts. A cancer paper mentioning coffee in passing lands in the cancer bucket. The counts measure how much research sits near a topic, not how much is squarely about it.

Counting is not appraising. A hundred small, poorly controlled human studies beat one excellent trial on this measure. Volume shows where attention went; it does not show who was right. For what the evidence actually concludes rather than how much of it exists, see is coffee healthy^[3].

About 64 papers match both the human and the animal pattern and are counted in both columns, so the two do not sum to the total.

The honest summary

Coffee is one of the most studied things people consume, and the studying is distributed very unevenly across the claims made for it. Exercise performance and caffeine itself rest on hundreds of randomised trials. Cancer, liver and the gut microbiome have far more papers than human evidence. The functional mushrooms added to coffee have, between them, fewer than forty papers reporting human participants across more than nine hundred studies.

When something is sold to you as backed by research, the useful questions are how much, in whom, and of what kind. For coffee, those answers are now countable, and they are on this page.