Mediterranean Diet

What it is

The Mediterranean diet is an eating pattern, not a single food or nutrient. The 2019 Cochrane review defined it as a high ratio of monounsaturated to saturated fat (with olive oil or tree nuts as the main fat sources), plenty of plant foods including fruit, vegetables and legumes, whole grains, more fish and less meat, and low to moderate wine as an optional extra [1]. In the PREDIMED trial, participants received extra virgin olive oil or mixed nuts on top of the diet [2]; in the Navarra sub-study the amounts were about 1 litre of olive oil a week or 30 g of nuts a day [3].

How it might work

Proposed, and not yet confirmed: the pattern's mix of unsaturated fats, fibre and plant compounds may lower inflammation and help blood vessels relax. In a meta-analysis of 57 trials, the diet improved markers including C-reactive protein, interleukin-6 and flow-mediated dilatation compared with control diets or usual care [4]. These are laboratory and blood-vessel markers, not heart attacks or deaths [4].

What the research says

The strongest evidence counts heart attacks and strokes themselves. In PREDIMED, 7,447 people at high heart risk but with no heart disease were assigned to a Mediterranean diet with extra virgin olive oil, a Mediterranean diet with nuts, or a low-fat diet; after a median 4.8 years, major heart events fell by about 30% in both Mediterranean groups (hazard ratios 0.69 and 0.72) [2]. The trial was withdrawn and re-published in 2018 after problems with how some participants were assigned, and the revised analysis confirmed the benefit [2]. CORDIOPREV then tested the diet in 1,002 people who already had heart disease and found fewer events over 7 years with the Mediterranean diet than with a low-fat diet (87 versus 111 events; hazard ratio 0.72 to 0.75 after adjustment) [5]. A 2024 meta-analysis of four large trials (10,054 people) found about half the odds of a major heart event with the Mediterranean diet (odds ratio 0.52), with smaller reductions in heart attacks and strokes; results varied a lot between trials [6].

Deaths are less clear-cut. A 2023 network meta-analysis of 40 trials (35,548 people), rated moderate certainty, found fewer deaths from any cause and from heart disease with Mediterranean-style programmes than with minimal advice [7]. The 2024 meta-analysis found fewer cardiovascular deaths (odds ratio 0.54) but no clear change in deaths from any cause [6]. A 2019 Cochrane review of 30 trials (12,461 people) rated the evidence low quality and found little or no effect on deaths in primary prevention, while strokes fell from about 24 to 14 per 1,000 people (moderate-quality evidence) [1]. Large observational reviews link higher adherence to about 23% lower all-cause mortality and 27% lower cardiovascular mortality in older adults, but they cannot prove cause [8].

Blood pressure falls a little. Meta-analyses of 35 trials (13,943 people) and of 19 trials (4,137 people) found average reductions of roughly 1-3 mm Hg [9][10]. Reductions were larger in people whose pressure started higher and in trials lasting 16 weeks or more [9]. Against other active diets, such as a low-fat diet, the difference was not significant [9]. In observational studies, higher adherence was linked to 13% lower odds of developing high blood pressure [10].

For weight, meta-analyses of dozens of randomised trials found slightly greater reductions in body weight and BMI with the Mediterranean diet than with comparison diets [4][11]. In adults with fatty liver disease, a 2025 meta-analysis of 11 trials found about 2.4 kg more weight loss than in controls [12]. For type 2 diabetes, a 2025 meta-analysis of 24 studies (about 992,000 people) found an 8% lower risk for each 2-point increase in adherence, with moderate certainty, and the one randomised trial it included pointed the same way [13]. PREDIMED also found less diabetes in its Mediterranean groups (hazard ratio 0.70) [14].

Cognitive health is less certain. A 2025 meta-analysis of 23 studies linked higher adherence to lower cognitive impairment, dementia and Alzheimer's disease, with significant heterogeneity between studies [15]. An earlier review of 34 prospective studies found less decline on global thinking tests but no significant link with dementia or mild cognitive impairment [16]. In a small PREDIMED sub-trial of 447 people, those on the Mediterranean diet scored better on several memory and thinking tests over about 4 years [3].

For cancer, observational meta-analyses covering more than 3 million people found about 13-14% lower cancer mortality and a lower risk of colorectal cancer, with weaker results for other cancers [17][18]. A breast-cancer analysis inside PREDIMED found fewer cases in the olive-oil group, but it was based on 35 cases and needs confirmation [19].

Doses used in studies

These are the exposures studied, not recommendations. PREDIMED supplied free extra virgin olive oil or mixed nuts alongside the diet and followed participants for a median 4.8 years [2]; in the Navarra sub-study the amounts were 1 litre of olive oil a week or 30 g of nuts a day [3]. CORDIOPREV compared a Mediterranean diet with a low-fat diet for 7 years in people with heart disease [5]. Blood-pressure trials that ran at least 16 weeks found larger reductions than shorter ones [9]. Most trials compared whole eating patterns rather than a set dose [1].

Safety and interactions

Side effects. Few trials report side effects in detail; where they did, adverse events were absent or minor [1].

Upper limits. As a whole eating pattern, the diet has no defined upper limit or toxic dose, and none is reported in the trials read for this page [1].

Medicines. The trials read for this page did not test interactions with medicines [1]; anyone taking medicines should ask a doctor before a big diet change.

Pregnancy. In a 2021 trial in 1,221 pregnancies at high risk of poor growth, the diet lowered the percentage of babies born small for gestational age (14% versus 22%) and a composite of pregnancy complications (18.6% versus 26.2%); the authors call these findings preliminary and in need of replication [20].

Alcohol. Some traditional versions include wine with meals, but the Cochrane definition treats wine as an optional addition [1]. Alcohol raises the risk of several cancers, and pooled research on light drinking found no safe level for cancer risk [21].

Other conditions. In fatty liver disease, trials found weight and liver-enzyme improvements [12]; the studies read for this page did not report results separately for kidney disease or bleeding disorders.

Caveats and open questions

Most of the evidence sits at low to moderate quality. A 2025 umbrella review of 18 meta-analyses (238 trials) found most were of low or critically low quality, and the effect was clearest in people who already had heart disease [22]. Trials disagree about deaths: one network meta-analysis found lower all-cause mortality, another pooled analysis found no clear effect [7][6]. PREDIMED was re-published after randomisation problems, and a Cochrane review remained cautious [2][1]. Findings for mortality, diabetes and dementia rely heavily on observational studies, which cannot show cause and effect [8][13][15]. How well the diet works in women is unclear: the CORDIOPREV benefit appeared in men, with no difference found in 175 women [5].