Four percent in a sealed chamber, forty grams on the scale#
Green tea has a measured effect on metabolism. Ten healthy men spent 24 hours in a respiration chamber on three occasions — taking green tea extract, matched caffeine, or placebo with each meal — and the extract raised 24-hour energy expenditure by 4 percent (P < 0.01) while lowering the respiratory quotient from 0.88 to 0.85 (P < 0.001), meaning proportionally more fat was being oxidised. Urinary norepinephrine excretion ran 40 percent higher than on placebo1. That is a coherent, mechanistically sensible, properly measured result.
Now the other end. Pooling randomised trials of at least 12 weeks in overweight and obese adults, the Cochrane review found that among the six trials conducted outside Japan — 532 participants — the mean difference in weight loss was −0.04 kg (95% CI −0.50 to 0.43; P = 0.88; I² = 18%)3. Forty grams, with a tight confidence interval and almost no heterogeneity. Not a noisy null — a confident one.
Both of those are careful studies. The distance between them is what this article is about: what the chamber measured, why it does not reach the scale, and the one genuine puzzle in the middle that nobody has solved. The broader case against shopping for burn-raising foods is in do 'metabolism-boosting' foods work; this is the flagship product on that shelf, taken apart.
Is it the catechins, or is it just the caffeine?#
This is the field's real disagreement, and it is worth stating precisely because both sides have a specific claim.
Dulloo's trial included an arm designed to answer exactly this. Alongside the green tea extract (50 mg caffeine plus 90 mg EGCG at each of three meals), participants took caffeine alone at the equivalent dose. That caffeine arm "had no effect on EE and RQ nor on urinary nitrogen or catecholamines"1. The conclusion drawn — and repeated ever since — is that catechins, not caffeine, drive the thermogenesis.
A later meta-analysis of respiration-chamber studies says otherwise, and says it with unusual clarity. Across six articles and 18 conditions, catechin-caffeine mixtures raised 24-hour energy expenditure by 428.0 kJ (4.7%, P < 0.001). Caffeine alone raised it by 429.1 kJ (4.8%, P < 0.001)2. One kilojoule apart. Whatever the extra ingredient is doing, it is not adding to the day's energy expenditure.
What separates the two results is dose and power, and it is nameable. Dulloo's caffeine-only arm delivered 150 mg across a whole day, in ten men — a modest dose at the bottom of the range where caffeine's own thermogenic effect is hardest to detect, in a sample small enough that a null is weak evidence of absence. The pooled chamber data spans higher doses and far more conditions. On the expenditure question, the meta-analysis is the better instrument, and it says the catechins are not the active part.
One difference does survive. On fat oxidation, catechin-caffeine mixtures produced a significant increase of 12.2 g/day (16.0%, P < 0.02) while caffeine alone reached 9.5 g/day (12.4%) but did not cross significance (P = 0.11)2. Read those carefully: the point estimates are close, and the gap between them is a difference in statistical significance, not a demonstrated difference in effect. It is the strongest surviving case for catechins doing something caffeine does not, and it is thin. What caffeine on its own is and is not doing is worked through in does caffeine actually burn calories.
| Outcome | Catechin + caffeine | Caffeine alone |
|---|---|---|
| 24-h energy expenditure | +428.0 kJ (4.7%), P < 0.001 | +429.1 kJ (4.8%), P < 0.001 |
| 24-h fat oxidation | +12.2 g (16.0%), P < 0.02 | +9.5 g (12.4%), P = 0.11 |
Why four percent never reaches the scale#
Four percent of a 2,400-calorie day is roughly a hundred calories, which sounds like it ought to do something over twelve weeks. Two things stop it.
The first is that a respiration chamber measures expenditure with intake fixed. Real people adjust. The second is habituation to caffeine, which is where most of the effect lives: the earlier meta-analysis of weight outcomes found the pooled effect shrinking sharply in people with high habitual caffeine intake — the analysis and its caveats sit in do 'metabolism-boosting' foods work. If you already drink coffee, the ingredient doing the work is one your body has largely stopped responding to.
A hundred extra calories a day, measured in a sealed room with the food weighed for you, does not survive contact with an ordinary week. The Cochrane estimate outside Japan is forty grams, and its confidence interval barely leaves zero in either direction.
Green tea also had no significant effect on the maintenance of weight loss3, which is the outcome that matters most to anyone who has already lost weight.
The Japanese trials, and the thing nobody can show#
The part of this literature that deserves genuine curiosity rather than dismissal is the geographic split, because it is stark. The Cochrane reviewers pooled the non-Japanese trials at −0.04 kg. The eight Japanese trials, covering 1,030 participants, reported weight reductions ranging from −0.2 kg to −3.5 kg. The reviewers analysed the two groups separately for an explicit reason: "due to the level of heterogeneity among studies"3.
What is striking is how thoroughly they failed to explain it. Dose was not the answer — "although the composition and dose of green tea was different among the studies, the range was not large and did not appear to be substantially different between the two subgroups." Baseline characteristics were not the answer either: "baseline characteristics of participants, including mean body weight and BMI, did not identify obvious sources of heterogeneity." Their summary is a shrug in formal dress: "the source of heterogeneity remains unclear; however, it appears to be associated primarily with studies conducted within Japan."
Two observations from the same review belong alongside that, and both cut in the same direction without proving anything. The reviewers noted that "there was, more often in studies conducted in Japan, a lack of detail provided as to how outcome measurements were made" — how weight was taken, where the waist was measured — which they flag as a route to inaccurate outcomes. And on funding, five of eighteen studies were judged at unclear risk of bias specifically because of how funding sources were reported or not reported; only one study in the entire review was rated low risk of bias across six domains.
That is as far as the evidence goes, and it is important not to push it further. Undisclosed funding is not proven industry influence, and weaker methods reporting is not proven bad measurement. What can be said is that the trials producing the impressive numbers are, as a group, the ones whose measurement procedures are least well described — and that a reviewer team with no stake in the answer looked hard for a benign explanation and could not find one.
The genetic hypothesis people reach for is real but unproven. Catechins inhibit catechol-O-methyltransferase, the enzyme that clears noradrenaline, and the COMT gene has common variants that differ in activity. A crossover pilot in 14 healthy Caucasian adults split by rs4680 genotype found that the high-activity (Val/Val) group responded to green tea with a rise in energy expenditure of 26.8 kJ over 3.5 hours and 3.0 g/day more fat oxidation, while the low-activity (Met/Met) group showed essentially no difference between green tea and placebo5. Seven people per genotype, and the authors call it a pilot. It is a lead, not an explanation — and 26.8 kJ is six calories.
The one number worth knowing, and it is about dose#
If you take nothing else from the practical end of this: the interesting threshold in green tea research is not a thermogenic one, it is a safety one, and it applies to extracts rather than to drinking tea.
EFSA's panel reviewed the evidence on green tea catechins and concluded that "intake of doses equal or above 800 mg EGCG/day taken as a food supplement has been shown to induce a statistically significant increase of serum transaminases" — a marker of liver injury. Catechins from green tea infusion prepared in the traditional way were, by contrast, judged generally safe at reported European intakes, which the panel put at a mean of 90 to 300 mg EGCG/day with high-level consumers reaching up to 866 mg/day4.
That pair of figures rewards careful reading rather than alarm. Drinking green tea is fine, including a lot of it. Concentrated extracts sold for weight loss are the form that reaches the concerning dose quickly, and they are also the form with the smallest demonstrated benefit — the Cochrane trials mostly used preparations, and still came out at forty grams. A supplement is where the risk is concentrated and the effect is not.
So the fair summary is short. Green tea genuinely raises energy expenditure by a few percent for a few hours; that rise appears to be caffeine's doing, and it fades if you are already a caffeine drinker; the fat-oxidation edge attributable to catechins is small and rests on a significance boundary; and in randomised trials outside Japan the whole thing amounts to nothing measurable on the scale. Drink it because you like it. The lever that moves weight is the size and consistency of the deficit — how big a calorie deficit should be — and no beverage is going to move a number set mostly by organs you cannot reach, which is the argument in metabolism explained.
FAQ#
How many cups of green tea would I need to drink to lose weight?#
There is no number that works, which is a more useful answer than a large one. Across six randomised trials outside Japan totalling 532 participants, the pooled weight difference against control was −0.04 kg with a confidence interval from −0.50 to 0.43 kg — a well-bounded zero after at least twelve weeks. The chamber studies that show a 4 percent rise in daily expenditure are measuring a real effect that does not survive into a free-living week.
Is green tea extract safe to take as a supplement?#
Ordinary brewed tea is generally regarded as safe, including at the higher end of European consumption. Concentrated extracts are the form that carries a documented risk: EFSA concluded that doses of 800 mg EGCG per day or above, taken as a food supplement, produce a statistically significant rise in serum transaminases, a liver-injury marker. Since the weight benefit in randomised trials is close to zero, an extract is the version of green tea with the worst ratio of risk to demonstrated effect. Anyone with liver concerns or on medication should raise it with a doctor before supplementing.
Why do studies from Japan find green tea works when others don't?#
Nobody has explained it, including the reviewers who looked hardest. Cochrane's team separated the two groups precisely because of heterogeneity, then checked the obvious candidates and ruled them out: catechin dose ranges were not substantially different between subgroups, and baseline weight and BMI did not account for the gap either. They did note that studies conducted in Japan more often failed to describe how outcomes were measured, and that five of eighteen studies were at unclear risk of bias over funding reporting. That is suggestive, not conclusive, and the honest position is that the split remains unexplained.
Sources#
- Dulloo AG, Duret C, Rohrer D, et al. Efficacy of a green tea extract rich in catechin polyphenols and caffeine in increasing 24-h energy expenditure and fat oxidation in humans. Am J Clin Nutr. 1999;70(6):1040-1045.
- Hursel R, Viechtbauer W, Dulloo AG, et al. The effects of catechin rich teas and caffeine on energy expenditure and fat oxidation: a meta-analysis. Obes Rev. 2011;12(7):e573-e581.
- Jurgens TM, Whelan AM, Killian L, Doucette S, Kirk S, Foy E. Green tea for weight loss and weight maintenance in overweight or obese adults. Cochrane Database Syst Rev. 2012;12(12):CD008650.
- EFSA Panel on Food Additives and Nutrient Sources added to Food (ANS), Younes M, et al. Scientific opinion on the safety of green tea catechins. EFSA Journal. 2018;16(4):e05239.
- Hursel R, Janssens PLHR, Bouwman FG, Mariman EC, Westerterp-Plantenga MS. The role of catechol-O-methyl transferase Val(108/158)Met polymorphism (rs4680) in the effect of green tea on resting energy expenditure and fat oxidation: a pilot study. PLoS One. 2014;9(9):e106220.
