Two middle-aged Finns can step on the same scale, register the same height, and produce an identical body mass index while carrying very different bodies underneath. A study of 2,264 adults published in the European Journal of Nutrition found that one thing separating them was how much coffee they drank.
Researchers at the University of Oulu analyzed participants in the Northern Finland Birth Cohort 1966, all measured at age 46 during a 2012 follow-up. People reporting the highest coffee intake had lower body fat percentage, lower total fat mass, less visceral fat, and more skeletal muscle than lighter drinkers. BMI, waist circumference, hip circumference, and waist-to-hip ratio were all comparable across the groups.
That combination is the story. Every crude measure of size said these people were the same. Every measure of composition said they were not.
The Difference BMI Was Never Going to Catch
Participants were sorted into four groups by self-reported daily intake: none, one to two cups, three to four cups, and five or more. Tea drinkers were excluded to keep the exposure clean. Finland is an unusual place to run this analysis because the heaviest category is also the largest one. Of 2,264 people, 1,076 reported five or more cups a day and only 70 reported none.
Body fat percentage fell across the groups at a p-value below .001. Fat mass followed at .002, and visceral fat area, the deep abdominal fat most closely tied to metabolic risk, at .048. Skeletal muscle mass moved the other way, rising with intake, again below .001. The pattern held when the researchers reran the analysis after excluding participants with extreme BMI values.
Coffee intake also correlated inversely with circulating branched-chain amino acids in both men and women. Chronically elevated levels of these amino acids have been linked in prior work to insulin resistance and to type 2 diabetes years before diagnosis. The authors call this a potentially novel finding for coffee specifically.
A Testosterone Signal That Points Two Ways at Once
The hormonal results are where the study becomes genuinely strange, and they showed up almost entirely in men, who made up 47% of the sample.
In fully adjusted models, each additional daily cup was associated with higher total testosterone, higher bioavailable testosterone, and higher sex hormone-binding globulin. At the same time, it was associated with lower free testosterone and a lower free androgen index. Total goes up, free goes down. The binding protein rising alongside the hormone explains the arithmetic.
Whether that is good or bad depends on which fraction matters. The authors point to a large analysis of U.S. men in which low total testosterone and low SHBG were each independently linked to metabolic syndrome, while calculated free and bioavailable testosterone were not. On that reading, the pattern coffee drinkers showed is the metabolically favorable one.
Women showed a narrower version: higher SHBG, lower free testosterone and a lower free androgen index, with no association for total or bioavailable testosterone. Women reporting polycystic ovary syndrome, or with no answer recorded, were excluded from those models.
"Coffee is consumed by millions of people every day, yet we still know surprisingly little about how it relates to our metabolism and hormones," said Luca Verroest, a doctoral researcher at the University of Oulu and the study's lead author.
Three Reasons to Hold This Loosely
The design is cross-sectional. Coffee habits and body composition were captured at the same visit, so nothing here establishes that one produced the other, and reverse causation is on the table. Leaner people with better glucose handling may simply tolerate more coffee.
Body composition was estimated by bioelectrical impedance rather than DXA or imaging. The authors flag this themselves, noting that impedance estimates rely on assumptions about body water and tissue conductivity. The visceral fat figure in particular is a derived estimate, not a scan.
Intake was self-reported through a single postal questionnaire that never defined a cup. No data were collected on bean type, roast, grind, or whether anyone added milk or sugar. Nearly all consumers reported filtered coffee, which matters because the diterpenes that raise LDL cholesterol are largely removed by a paper filter. A study of espresso drinkers might not look the same.
The correlations themselves were small, generally under 0.15 across the metabolite panel and under 0.2 for cardiometabolic markers. Testosterone also varies substantially within a single person from week to week, and it was measured once. The authors are explicit that these are population-level patterns rather than changes anyone would detect in their own bloodwork.
Higher coffee consumers in this cohort were also more likely to smoke and less likely to hold a university degree, though the regression models adjusted for both.
What Would Settle It
The Oulu team says the next step is determining whether coffee itself drives any of this and which of its roughly one thousand bioactive compounds might be responsible. Caffeine is the obvious candidate, and genetic analyses have tied higher predicted plasma caffeine to lower fat mass. But one trial in trained athletes found decaffeinated coffee also raised post-exercise testosterone, which points elsewhere. Chlorogenic acids and trigonelline are among the alternatives.
That work is running in animal models now, with human intervention trials the eventual goal. Until those exist, nothing in this paper supports drinking more coffee to change body composition, and existing federal guidance on caffeine has not moved. The FDA still puts 400 milligrams a day, roughly four or five cups, as the level not generally associated with negative effects in healthy adults. People who are pregnant, managing a heart rhythm disorder, or taking medications affected by caffeine should raise the question with a clinician rather than a coffee study.
Key Questions Answered
What did the study actually measure?
Coffee intake by questionnaire, body composition by bioelectrical impedance, and blood markers including metabolites, glucose, insulin, and sex hormones, all at a single visit when participants were 46.
Does this mean coffee burns visceral fat?
No. The design cannot separate cause from effect. Leaner, more metabolically healthy people may simply drink more coffee.
Why did total testosterone rise while free testosterone fell?
Sex hormone-binding globulin rose alongside total testosterone. More of the hormone was bound to that protein, leaving a smaller unbound fraction.
Do the findings apply to women?
Partly. Women showed higher SHBG and lower free androgen measures, but no association with total or bioavailable testosterone.
How much coffee is considered safe?
The FDA cites 400 milligrams of caffeine daily, roughly four to five cups, as not generally associated with negative effects in healthy adults. Individual tolerance varies.
Should anyone change their coffee habit based on this?
No. The researchers say longitudinal and intervention studies are needed before any of this becomes dietary guidance.
