Which diet is best? What the studies say
In short: no diet is best for everyone, but they do not all rest on the same evidence. The Mediterranean diet has the strongest support: large clinical trials show about 25 to 30% fewer cardiovascular events. DASH lowers blood pressure, vegetarian diets lower weight and cholesterol, and keto melts liver fat in the short term. The carnivore diet has no serious study behind it. What to do? Choose according to your problem (heart, blood pressure, weight, liver) and to what you can keep up for years, and talk to your doctor if you take medication.
Updated on 13 November 2026.
Every year brings a new miracle diet, and every diet has converts who swear everyone else is wrong. Gil Carvalho, a physician and researcher who runs the YouTube channel Nutrition Made Simple!, published a video at the end of October 2026 ranking the most popular diets on three criteria: their measured effects, the strength of that evidence, and how easy they are to follow in real life. We took his ranking and checked each of his claims against the original source. Verdict: he is right almost everywhere, with two caveats that we point out.
How do you judge a diet?
Not all studies are equal. To judge a diet, three questions matter.
What kind of study? In a randomised trial, participants are assigned to a diet by lot: it is the only way to prove that a diet causes an effect. An observational study follows people who eat as they please: it shows an association, not a cause. Vegetarians, for example, smoke less and exercise more than average, which muddies comparisons.
What outcome? A drop in cholesterol or blood pressure is a marker: useful, but indirect. The outcomes that really count are hard events: heart attacks, strokes, cancers, deaths. Few diets have been tested on these, because it takes thousands of people followed for years.
For how long? Almost every diet produces weight loss in the first months. What sets a good diet apart is what remains after two, five or ten years, and whether people manage to stick with it that long.
The ranking in one table
| Diet | Principle | Strongest evidence | Level of evidence | G. Carvalho's score |
|---|---|---|---|---|
| Mediterranean | Vegetables, fruit, legumes, whole grains, olive oil, nuts, fish; little red meat | Fewer heart attacks and strokes (3 long trials) | High (long randomised trials) | 8/10 |
| DASH | Close to Mediterranean, rich in potassium, low in salt | Lower blood pressure | High for blood pressure, low for heart events | 6.5/10 |
| Pescatarian | Mediterranean without meat, with fish | Fewer deaths in observational studies | Low to moderate (no long trial) | 6/10 |
| Vegetarian / vegan | No meat or fish (vegan: no animal products at all) | Lower weight and cholesterol; fewer cancers by association | Moderate (short trials + observational) | 6 to 7/10 |
| Low-fat | Less than 20 to 30% of calories from fat | Fewer deaths compared with usual diet | Moderate | 5 to 5.5/10 |
| Paleo | Meat, fish, vegetables, fruit; no grains, legumes or dairy | Short-term weight loss and lower triglycerides | Low (no long-term data) | 5 to 5.5/10 |
| Ketogenic (keto) | Less than 10% of calories from carbohydrates | Liver fat, weight, blood sugar in the short term; epilepsy | Moderate short term, none long term | 5 to 6/10 |
| Carnivore | Animal products only | No controlled study | Very low (online survey) | 2 to 3/10 |
Gil Carvalho's score blends evidence, health effects and ease of adherence. A diet can get an average score not because it is bad, but because long-term data are missing.
The Mediterranean diet: the best supported
The "Mediterranean diet" of scientists is not the actual cooking of Southern Europe, which is often rich in cured meats and white bread. It is a model: plenty of vegetables, fruit, legumes, whole grains, nuts and olive oil, fish and poultry, fermented dairy, and little red meat, sugar or ultra-processed food.
Its strength: three randomised trials lasting several years on heart disease (Lyon in 1999, PREDIMED, CORDIOPREV), all positive. The latest, CORDIOPREV (Spain, 2022), followed 1,002 people who had already had a heart problem for seven years. Compared with a low-fat diet, the Mediterranean diet reduced heart attacks, strokes and other cardiovascular events by about 25 to 28%, depending on the analysis. The 28% figure quoted in the video is correct; it is the most favourable value. The effect was clear in men; in women, who were few (under 20% of participants), it was not significant, which does not prove it was absent. And these results concern people who already had heart disease, not the general population.
In the same trial, the thickness of the carotid artery wall (the arteries in the neck), a marker of atherosclerosis, decreased on the Mediterranean diet and not on the low-fat diet. The decrease is small (about 0.03 mm), but it goes in the right direction.
In 2023, a large review in the BMJ (40 trials, more than 35,000 participants at cardiovascular risk) concluded that the Mediterranean diet lowers mortality by about 28% compared with "minimal intervention", meaning the usual, often unhealthy, diet. Certainty: moderate.
The limits. The Mediterranean diet depends on what is actually on the plate: pouring olive oil on a hot dog does not make it Mediterranean. You can also gain weight on it, as oil and nuts are high in calories. Its great advantage remains that it is the least restrictive on the list: nothing is forbidden, which helps people stick with it for years.
The DASH diet: made for blood pressure
DASH stands for Dietary Approaches to Stop Hypertension. It is a cousin of the Mediterranean diet, with two emphases: plenty of potassium-rich foods (vegetables, fruit, legumes, low-fat dairy) and little salt.
The founding trial of 1997 showed a drop in blood pressure of about 5.5/3 mmHg in eight weeks, and 11/5.5 mmHg in people with hypertension: in the range of what a medicine achieves. This does not replace treatment when a doctor considers it necessary. Cutting salt as well strengthens the effect.
The video also cites a trial in which "soft" plaque in the coronary arteries (non-calcified, the type most likely to rupture) shrank more with DASH. A caveat: in the small Polish DISCO-CT trial (fewer than 100 people), the DASH diet was part of a programme that also added more physical activity, so the effect cannot be attributed to the diet alone. And this result was not the trial's main outcome.
What is missing: large long-term trials on heart attacks and deaths, as for the Mediterranean diet. DASH is not necessarily less healthy; it is less studied. For more on measuring blood pressure, see our article Checking your blood pressure: why and how?.
The low-fat diet: not so outdated
Very fashionable in the 1990s, the low-fat diet is now seen as outdated. Yet the data are better than people think: in the 2023 BMJ review, it also lowered mortality compared with the usual diet, with no clear difference from the Mediterranean diet. This result becomes less solid, however, when trials that combined the diet with stopping smoking or with medication are removed.
The large American Women's Health Initiative trial (nearly 49,000 women, eight years) aimed for a sharp cut in fat. Participants did not manage it: the difference achieved was small, and no drop in cardiovascular disease was seen. That is the main weakness of this diet: it is hard to stick to. And by cutting out all fats, you also lose the good ones (olive oil, nuts, oily fish). On this point, see our article Fats and cholesterol: what has really changed.
Pescatarian, vegetarian, vegan
The pescatarian diet is a Mediterranean diet without meat but with fish, a good source of protein, omega-3, vitamin B12 and iodine. Observational studies associate it with lower mortality than among meat eaters, but no long trial has tested it.
Vegetarian (no meat or fish, with eggs and dairy) and vegan (no animal products at all) diets have been studied a lot over short periods. A review of 11 randomised trials in people who were overweight or diabetic shows, with a vegan diet, a loss of about 4 kg, a small drop in LDL cholesterol and in glycated haemoglobin (the diabetes marker), but no effect on blood pressure. Against another active diet, the weight difference shrinks to 2.7 kg. A trial in 22 pairs of identical twins (2023, eight weeks) confirmed a larger drop in LDL with a healthy vegan diet than with a healthy omnivorous diet.
Cancer. A 2026 meta-analysis (17 publications) associates vegetarian diets with about 13% fewer cancers overall, and with lower risk for several cancers: stomach, colon and rectum, pancreas, breast, bladder, melanoma, lymphoma. Among vegans, only total risk is clearly lower. But a large analysis published the same year (1.8 million people, nine studies) qualifies this picture: vegetarians had fewer cancers of the pancreas, breast, prostate and kidney, but not fewer colorectal cancers, and vegans even had a higher risk of colorectal cancer (+40%), possibly linked to a lower intake of certain nutrients such as calcium. These are observational studies: they do not prove that the diet itself protects or harms, and vegans are few in them. The video does not mention this second result.
Points to watch, especially for vegans:
- vitamin B12: a supplement is essential;
- vitamin D, calcium and iodine: to monitor (fortified plant drinks, iodised salt);
- fractures: a higher risk was observed among vegans in the United Kingdom, probably linked to lower calcium and protein intake and lower body weight;
- iron and long-chain omega-3: to monitor case by case.
Protein, on the other hand, is rarely a problem with a varied diet (legumes, soy, grains). And "vegan" does not mean "healthy": crisps and soft drinks are vegan. It all depends on what you eat, not only on what you avoid.
Paleo, keto, carnivore: the low-carb diets
These three diets cut carbohydrates (sugars and starches), more and more strongly.
The paleo diet imitates what prehistoric humans supposedly ate: meat, fish, vegetables, fruit, nuts, without grains, legumes or dairy. The irony: our ancestors did eat wild grains and legumes. Trials show weight loss and lower triglycerides, clear in the first months and fading afterwards. The longest, in Sweden, lasted two years in postmenopausal women. There are no data on heart attacks, cancers or deaths.
The ketogenic (keto) diet limits carbohydrates to less than 10% of calories, which makes the liver produce ketone bodies, a back-up fuel. It has a recognised medical use: reducing seizures in some children with epilepsy that medicines do not control. Otherwise, trials show weight loss and lower blood sugar and triglycerides, mostly in the short term. LDL cholesterol rises in some people, depending on the type of fat eaten.
The most striking result is recent. In August 2026, a team at Washington University in St. Louis compared three diets in 42 people with obesity, prediabetes and fatty liver (steatosis). All meals were provided, and everyone lost about 10% of their weight. Liver fat fell by 67% on keto, against 45% on the Mediterranean diet and on a very low-fat diet. Half of the keto group were no longer prediabetic, against 29% and 7% in the other two groups. The video's figure is therefore correct, but the trial is small (14 people per group) and short (four to five months). Nobody knows whether the advantage lasts.
The limits of keto. No long trial shows what happens after ten years. Large observational studies on low-carb diets (not strictly keto) find lower mortality when carbohydrates are replaced by plant foods (nuts, oils, legumes), and higher mortality when they are replaced by animal products. The keto popularised on social media, based on bacon, butter and coconut oil, has little to do with a keto built on olive oil, nuts, salmon and green vegetables. And most people find it hard to stick to.
The carnivore diet cuts out all plant foods. There is no controlled trial. The main published study is a survey shared on social media in 2020 among 2,029 followers, who said they were in good health. Their answers were not verified, and those who did badly on the diet had probably stopped before the survey. Among those who reported their blood tests, LDL cholesterol was clearly high (172 mg/dl on average). Long-term risks have never been measured, but the concerns are serious: the heart (high LDL), the colon (lots of red meat, no fibre, frequent constipation), the gut microbiome, and possible deficiencies.
True or false: common beliefs
"The Mediterranean diet is mostly about olive oil." False. Olive oil matters, but it is the whole pattern that protects: vegetables, legumes, whole grains, nuts, fish, and little red meat and ultra-processed food.
"Keto is the most effective diet for losing weight." False in the long run. It often leads to faster loss at first, but after a year or two the differences between diets shrink considerably. The best diet for losing weight is the one you can stick to.
"Vegetarians lack protein." Mostly false. With legumes, soy, grains, eggs and dairy, intake is sufficient, even for building muscle. Vitamin B12, on the other hand, really is lacking in vegans who take no supplement.
"The low-fat diet is outdated." It depends. It lowers mortality compared with the usual diet, but it is hard to stick to and also removes the good fats.
"The carnivore diet cures autoimmune diseases." Not proven. Testimonials exist, but a diet that removes almost all foods also removes those that are poorly tolerated, which may bring relief without treating the cause.
Which diet for which goal?
| Your situation | Best-supported diet | Comment |
|---|---|---|
| Previous heart attack or stroke | Mediterranean | Best supported by long trials in these patients |
| High blood pressure | DASH, with less salt | Effect in the range of a medicine, within weeks |
| High LDL cholesterol | Mediterranean, vegetarian or vegan | Replace saturated fats with unsaturated ones |
| Overweight | The one you will stick to | Differences between diets fade after a year |
| Fatty liver (steatosis), prediabetes | Mediterranean; keto under medical supervision | Keto did better in the short term in a small trial |
| Drug-resistant epilepsy | Keto, in a specialist setting | Recognised medical use |
| Wish to eat no animal products | Well-planned vegetarian or vegan | B12 supplement essential for vegans |
These diets are not mutually exclusive: you can follow a low-carb Mediterranean diet, or a vegan keto diet.
In practice
- Start from the Belgian guidelines. The Belgian Superior Health Council recommends, among other things, limiting red meat to 300 g per week and processed red meat (cold cuts) to 30 g per week, eating legumes several times a week and a small handful of unsalted nuts or seeds every day (20 to 30 g). In practice, that is a Mediterranean-style diet.
- Choose according to your problem (table above), not according to fashion.
- Think long term. A diet that is perfect on paper but dropped after three months protects against nothing.
- Judge by measurements. Before and a few months after a major change: weight, waist size, blood pressure, blood test (cholesterol, blood sugar, liver enzymes).
- See your doctor before a very restrictive diet (keto, carnivore, strict vegan) if you are diabetic, pregnant, elderly, on medication, or have had an eating disorder.
- Avoid diets sold with products. A diet does not need branded supplements, except vitamin B12 for vegans.
For more on sugar, see our article Am I eating too much sugar?, and on the role of food in inflammation, Inflammation: vital ally or silent enemy?.
Gil Carvalho's video
Quiz
Further reading
Hazard ratio and relative reduction. When a trial reports "28% fewer events", that is a relative reduction. In CORDIOPREV, about 17% of patients in the Mediterranean group and 22% in the low-fat group had an event over seven years. The absolute reduction is therefore about 5 points: around twenty patients had to follow the diet for seven years to prevent one event. That is a lot for a simple change of diet.
Compared with what? CORDIOPREV's "low-fat" diet was not that low in fat: about 32% of calories from fat, against 40% for the Mediterranean diet. And in the BMJ review, both diets are mainly compared with the usual diet. A diet that beats "nothing at all" is not necessarily better than another good diet.
The healthy-adopter bias. In observational studies, people who choose a vegetarian or Mediterranean diet often have other healthy habits. Researchers adjust for smoking, exercise and income, but never perfectly. Hence the importance of randomised trials, which are rare and costly in nutrition.
Fat quality. The keto trials with favourable results often use unsaturated fats. Gil Carvalho cites a very small trial comparing two keto diets identical except for the type of fat, which found better results (insulin, blood sugar, LDL) with polyunsaturated fats (nuts, fish) than with saturated fats (butter, fatty meat), which matches everything else we know.
Sources
- Appel, L. J., Moore, T. J., Obarzanek, E., et al. (1997). A clinical trial of the effects of dietary patterns on blood pressure. New England Journal of Medicine, 336(16), 1117-1124. https://doi.org/10.1056/NEJM199704173361601
- Aune, D., Schlesinger, S., & Sobiecki, J. G. (2026). Vegetarian and vegan diets and cancer incidence: a systematic review and meta-analysis of prospective studies. European Journal of Epidemiology, 41, 531–545. https://doi.org/10.1007/s10654-026-01380-8
- Carvalho, G. [Nutrition Made Simple!]. (2026, 28 October). I Studied Nutrition for 30 Years. Here's How I Rank Every Popular Diet [Video]. YouTube. https://youtu.be/HO3EtjISxgM
- de Lorgeril, M., Salen, P., Martin, J.-L., et al. (1999). Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study. Circulation, 99(6), 779-785. https://doi.org/10.1161/01.CIR.99.6.779
- Delgado-Lista, J., Alcala-Diaz, J. F., Torres-Peña, J. D., et al. (2022). Long-term secondary prevention of cardiovascular disease with a Mediterranean diet and a low-fat diet (CORDIOPREV): a randomised controlled trial. The Lancet, 399(10338), 1876-1885. https://doi.org/10.1016/S0140-6736(22)00122-2
- Dunneram, Y., Lee, J. Y., Watling, C. Z., et al. (2026). Vegetarian diets and cancer risk: pooled analysis of 1.8 million women and men in nine prospective studies on three continents. British Journal of Cancer, 134, 1218-1229. https://doi.org/10.1038/s41416-025-03327-4
- Estruch, R., Ros, E., Salas-Salvadó, J., et al. (2018). Primary prevention of cardiovascular disease with a Mediterranean diet supplemented with extra-virgin olive oil or nuts. New England Journal of Medicine, 378(25), e34. https://doi.org/10.1056/NEJMoa1800389
- Henzel, J., Kępka, C., Kruk, M., et al. (2021). High-risk coronary plaque regression after intensive lifestyle intervention in nonobstructive coronary disease: a randomized study. JACC: Cardiovascular Imaging, 14(6), 1192-1202. https://doi.org/10.1016/j.jcmg.2020.10.019
- Howard, B. V., Van Horn, L., Hsia, J., et al. (2006). Low-fat dietary pattern and risk of cardiovascular disease: the Women's Health Initiative randomized controlled dietary modification trial. JAMA, 295(6), 655-666. https://doi.org/10.1001/jama.295.6.655
- Jiménez-Torres, J., Alcalá-Diaz, J. F., Torres-Peña, J. D., et al. (2021). Mediterranean diet reduces atherosclerosis progression in coronary heart disease: an analysis of the CORDIOPREV randomized controlled trial. Stroke, 52(11), 3440-3449. https://doi.org/10.1161/STROKEAHA.120.033214
- Karam, G., Agarwal, A., Sadeghirad, B., et al. (2023). Comparison of seven popular structured dietary programmes and risk of mortality and major cardiovascular events in patients at increased cardiovascular risk: systematic review and network meta-analysis. BMJ, 380, e072003. https://doi.org/10.1136/bmj-2022-072003
- Landry, M. J., Ward, C. P., Cunanan, K. M., et al. (2023). Cardiometabolic effects of omnivorous vs vegan diets in identical twins: a randomized clinical trial. JAMA Network Open, 6(11), e2344457. https://doi.org/10.1001/jamanetworkopen.2023.44457
- Lennerz, B. S., Mey, J. T., Henn, O. H., & Ludwig, D. S. (2021). Behavioral characteristics and self-reported health status among 2029 adults consuming a "carnivore diet". Current Developments in Nutrition, 5(12), nzab133. https://doi.org/10.1093/cdn/nzab133
- Martin-McGill, K. J., Bresnahan, R., Levy, R. G., & Cooper, P. N. (2020). Ketogenic diets for drug-resistant epilepsy. Cochrane Database of Systematic Reviews, 6, CD001903. https://doi.org/10.1002/14651858.CD001903.pub5
- Mellberg, C., Sandberg, S., Ryberg, M., et al. (2014). Long-term effects of a Palaeolithic-type diet in obese postmenopausal women: a 2-year randomized trial. European Journal of Clinical Nutrition, 68(3), 350-357. https://doi.org/10.1038/ejcn.2013.290
- Ornish, D., Scherwitz, L. W., Billings, J. H., et al. (1998). Intensive lifestyle changes for reversal of coronary heart disease. JAMA, 280(23), 2001-2007. https://doi.org/10.1001/jama.280.23.2001
- Petersen, M. C., Smith, G. I., Farabi, S. S., et al. (2026). Effect of diet macronutrient content on the cardiometabolic response to weight loss: a randomized clinical trial. Cell Metabolism. https://doi.org/10.1016/j.cmet.2026.07.020
- Seidelmann, S. B., Claggett, B., Cheng, S., et al. (2018). Dietary carbohydrate intake and mortality: a prospective cohort study and meta-analysis. The Lancet Public Health, 3(9), e419-e428. https://doi.org/10.1016/S2468-2667(18)30135-X
- Superior Health Council. (2019). Food-based dietary guidelines for the Belgian adult population (advisory report no. 9284). FPS Public Health. https://www.hgr-css.be
- Superior Health Council. (2025). Food-based dietary guidelines for the Belgian population (advisory report no. 9805-9807). https://www.hgr-css.be
- Termannsen, A.-D., Clemmensen, K. K. B., Thomsen, J. M., et al. (2022). Effects of vegan diets on cardiometabolic health: a systematic review and meta-analysis of randomized controlled trials. Obesity Reviews, 23(9), e13462. https://doi.org/10.1111/obr.13462
- Tong, T. Y. N., Appleby, P. N., Armstrong, M. E. G., et al. (2020). Vegetarian and vegan diets and risks of total and site-specific fractures: results from the prospective EPIC-Oxford study. BMC Medicine, 18, 353. https://doi.org/10.1186/s12916-020-01815-3