Best Supplements for Weight Loss: Nothing Here Is Established for Weight Loss. The Two Best-Evidenced Answers Are Food.
The category has been reviewed, and the review could not answer the question. It pooled 21 studies of products sold to burn fat, and in every one the people taking them were also dieting, exercising or both; its authors write that comparing supplement with no supplement was not possible. [the full verdict]
We have since checked 19 of the things sold for weight loss, one at a time, reading the trials instead of the labels. This page is the running scoreboard, and it has an awkward shape.
Not one entry on it is rated Established for weight loss. The two with the best evidence behind them are fiber and protein — neither marketed as a fat burner, both of them food.
Nothing here is for sale and nothing links to a shop. Every row goes to the verdict it came from, so you can check the working.
| What | Verdict | What we measured |
| Psyllium husk The only item near this topic with settled evidence behind it, and what it is settled for is cholesterol. Its own verdict is titled not nature’s Ozempic, which tells you how the weight-loss claim went. | established | 13 mg/dL off LDL — the claim that actually holds up |
| What | Verdict | What we measured |
| Fiber It works, and the average is easy to misread — which is most of the verdict. | supported | 2.8 lb average extra loss, 22 supplement trials |
| Protein, for fat loss It works. For most dieters, your target is probably about a third higher than it needs to be. | supported | 0.9 lb of lean mass kept |
| Protein, for fullness Yes — and in the trial that ranked foods by how full they left people, a boiled potato beat everything. | supported | 441 calories a day, eaten without noticing, by 19 adults — 16 of them women |
| Chewing gum About 11 calories an hour, measured in seven people chewing for twelve minutes. It is here because it is real, and a useful sense of scale for everything above it. | supported | 11 extra calories an hour while chewing, measured in seven people |
| What | Verdict | What we measured |
| Drinking water A little, and not by the mechanism everyone repeats. | preliminary | 24 calories, the famous 30% boost |
| Oatmeal Maybe a little. Pooled trials found about 2 lb at best, and most did not test a bowl of oatmeal. What regulators signed off on is cholesterol. | preliminary | 75 glycaemic index of instant oatmeal — large-flake is 53 |
| Apple cider vinegar About 2 lb across twelve short trials, and in the one that checked, most of it came back after stopping. The headline trial was retracted. | preliminary | 79 % of the weight lost that returned within four weeks of stopping |
| Akkermansia Eleven small trials, split. The largest missed its main goal: 0.15 lb from a dummy capsule after four months. | preliminary | 0.15 lb difference in weight from a dummy capsule after four months of pasteurized Akkermansia, in the largest trial, 142 adults (0.07 kg); the trial missed its main goal |
| What | Verdict | What we measured |
| Colostrum Rated Preliminary, for a gut claim. On body fat, eight controlled trials set it against another protein powder and it came out ahead in none. | preliminary | 5 small trials in healthy volunteers behind the gut result; in one of them the marker rose on colostrum |
| Greens powders Rated Preliminary, for energy, immunity and digestion. The vitamins reach your blood; the biggest brand’s gut result missed twice. Weight was never rated. | preliminary | 30 people in the biggest brand’s 2024 trial, registered two months after it ended |
| What | Verdict | What we measured |
| Fat burners 21 studies, and not one tested the pill without diet or exercise. | unsupported | 21 studies. not one without diet or exercise |
| Coffee Ten trials handed people real coffee. Pooled, nothing you could call an effect. | unsupported | 1.4 lb coffee against a placebo drink, and no effect is inside the range |
| Berberine “Nature’s Ozempic” is 1.9 pounds. The drug beat placebo by 28. | unsupported | 14x more weight lost on the drug than on the supplement named after it, each against people not taking it |
| Creatine, for fat loss No — and it moves your scale the wrong way while not doing it. | unsupported | 1.9 lb added to your scale mid-diet — and almost none of it is fat |
| Detox teas Nobody selling one will name the toxin. | unsupported | 0 randomised trials of a tea sold for detox or weight loss, 2002 to 2024 |
| Lemon water No. It does not alkalise you either. | unsupported | 0 alkalising effect of lemon on urine pH, measured |
| Negative calorie foods One small study put celery 86% of the way there, and it could not tell nearly free from break-even. | unsupported | 86% of celery’s calories went on digesting it, on average, in one study of 15 women |
| Sea moss The NIH’s own database says no medicinal value has been proven. | unsupported | 57 adults in a placebo trial that found no edge |
Established means multiple good studies agree: treat it as settled. Supported means the evidence is consistent, with known limits. Preliminary means the evidence is early, small, conflicting or from animals only: not enough to act on alone. Unsupported means we looked and the evidence does not back the claim: either it was tested and did not hold up, or nobody has tested it and nothing that was tested supports it.
Read the shape of this page rather than the rows. The top section contains nothing that burns fat. The bottom section is where the marketing budget went. And the two entries with the best evidence behind them are fiber and protein, both of them food. That is not a coincidence — it is what a category looks like when it is sold on a promise the trials were never asked to test.
The same corpus, asked other ways: for muscle growth, for men and for women, where the question becomes who was actually in the trials, for gut health, where the winner is fiber again, and for recovery, where the best-evidenced item is also not a supplement, and for sleep, where exactly one entry is Supported and what it buys is measured in minutes, and for sweeteners, where every claim is a worry and the ratings run backwards.
The four ratings are the same ones used on every verdict here, and we wrote down how we read a study so you can see what earns each one.