The diets with the cleanest longevity evidence all converge on the same plate: lots of vegetables, legumes, nuts, fish, whole grains, and olive oil — and not much else. Get that pattern roughly right and the differences between named diets are small.

Dietary patterns

In nutrition science, dietary patterns consistently outperform individual nutrients or supplements. The Mediterranean, MIND (a Mediterranean–DASH hybrid emphasising leafy greens and berries for brain health), and DASH (Dietary Approaches to Stop Hypertension) diets, along with the eating patterns of long-lived Blue Zone populations, share enough common ground that the practical recommendations converge. The Mediterranean pattern has the deepest evidence base of the named diets — a 2024 review of 28 studies in older adults found roughly 23% lower all-cause mortality and 27% lower cardiovascular mortality with high adherence, though non-fatal cardiovascular events fell short of significance.^[1]

The Mediterranean diet: the strongest evidence base

The single most-validated dietary intervention in longevity science.

Core components

Vegetables and fruit — 5+ servings/day
Whole grains — bread, pasta, rice, oats (whole-grain forms)
Legumes — 3+ servings/week (beans, lentils, chickpeas)
Nuts and seeds — daily handful
Extra-virgin olive oil — primary cooking fat (at least 4 tablespoons/day, ~50 g, in the PREDIMED protocol)
Fish and seafood — 2+ servings/week, especially fatty fish
Modest dairy — primarily yogurt and cheese
Moderate poultry and eggs — several times/week
Limited red meat — once/week or less
Limited sweets and processed food
Wine in moderation (Mediterranean tradition; the modern evidence on alcohol has shifted — see Alcohol)

Hard outcome data

PREDIMED — a landmark Spanish trial of 7,447 adults at high cardiovascular risk found roughly 30% fewer heart attacks, strokes and cardiovascular deaths (HR ~0.70) on the Mediterranean diet versus a low-fat control. Moderate. It remains one of the few large randomised nutrition trials with hard endpoints, but with an important caveat: the original 2013 paper was retracted and republished in 2018 because of randomisation irregularities at some sites (household- and clinic-level assignment rather than clean individual randomisation). The ~30% relative reduction survived re-analysis, so the signal holds — but it is no longer a cleanly individually-randomised trial.^[2] The olive oil and nuts were donated by the Spanish olive and nut industries.
A 2024 review of 28 studies in older adults (mean ages 64.8–69.8) found high Mediterranean adherence associated with about 23% lower all-cause mortality (pooled relative risk 0.77, 95% CI 0.70–0.83, from 19 studies and 158,520 people) and about 27% lower cardiovascular mortality (0.73, 0.64–0.84, from 9 studies and 151,170 people). Non-fatal cardiovascular events pointed the same way but were not statistically significant — 0.75, with an interval of 0.55 to 1.01 that includes no effect at all, from 7 studies and 13,264 people. The nearly 680,000 figure often quoted from this review is the total across all 28 studies, not the base of any single estimate. Of its two randomised studies, both are PREDIMED, so the randomised half adds no independent trial beyond the one described above.^[3]
PREDIMED-Plus tests an energy-restricted Mediterranean diet for weight loss in adults with cardiometabolic risk. Its 1-year data show a mean weight difference of −2.5 kg versus control (95% CI −3.1 to −1.9 — the bracketed range is where the true effect most plausibly lies; 33.7% vs 11.9% achieving ≥5% loss), but its primary cardiovascular endpoint — a composite of cardiovascular death, nonfatal heart attack and nonfatal stroke — is still unpublished.^[4] One hard disease endpoint is now in, though: over a median six years the energy-restricted arm had 31% fewer new cases of type 2 diabetes (95% CI 18% to 41%), an absolute fall from 12.0% to 9.5%.^[5] Read that as evidence for adding calorie restriction and exercise on top of a Mediterranean diet, not for the diet against a non-Mediterranean comparator — both arms were given Mediterranean-diet advice.
Pooling 54 cohorts, 1.8 million people and 346,034 deaths, each one-point rise in Mediterranean adherence score tracked with about 4% lower all-cause mortality (relative risk 0.96, 95% CI 0.95–0.97), rated moderate certainty — which compounded over the four-to-five-point gap between high and low adherence works out to roughly 15–19%, close to what individual cohorts report.^[6] Moderate. Recent specific anchors: in the Women's Health Study (25,315 women, mean 24.7 years, 3,879 deaths), the highest adherence group had 23% lower all-cause mortality adjusting only for age, treatment and energy intake (hazard ratio 0.77, 95% CI 0.70–0.84) — but 11% once smoking, physical activity, alcohol and menopausal factors were added (0.89, 0.82–0.98). Much of the raw association is the rest of the lifestyle that travels with the diet. The largest of a small set of measured contributors to what remained were small-molecule metabolites, inflammation, body-mass index and triglyceride-rich lipoproteins rather than standard cholesterol or glucose measures — though all 33 biomarkers together account for only about a fifth of the association (21.3%), leaving four fifths unexplained by any measured cardiometabolic pathway.^[7]
A 2025 meta-analysis (23 studies) found strong Mediterranean adherence associated with an 11–30% lower risk of age-related cognitive disorders — pooled hazard ratios of 0.82 for cognitive impairment (about 18% lower), 0.89 for dementia (11% lower), and 0.70 for Alzheimer's disease (30% lower). Those three figures are three different outcomes in three different study pools, not one range — and they are contested.^[8] A 2026 meta-analysis over an overlapping literature found the Mediterranean–dementia association not statistically significant (hazard ratio 0.85, 95% CI 0.71–1.03), with only MIND reaching significance (0.78, 0.68–0.89); the Mediterranean signal appeared only in studies with follow-up under ten years, which is the signature of early cognitive decline changing what people eat rather than the reverse.^[9] The standing Lancet Commission on dementia prevention lists 14 modifiable risk factors and no dietary pattern among them.^[10] Moderate.
Mediterranean adherence is associated with lower type 2 diabetes, and this is the one place the evidence is randomised rather than observational.^[11] Cancer is more selective than a blanket "lower cancer incidence" suggests: pooling 117 studies and 3.2 million people, adherence tracked with lower colorectal (relative risk 0.83), gastric (0.70), liver (0.64) and breast (0.94) cancer, but made no difference to blood, oesophageal, pancreatic or prostate cancer, and the review reports no overall cancer-incidence figure at all. The steadier signal is cancer mortality, about 13% lower (0.87, 95% CI 0.82–0.92).^[12]

What's actually doing the work

The Mediterranean pattern's success cannot be pinned to a single nutrient. Three mechanisms have the strongest support:

1. Polyphenols, not just "healthy fat." PREDIMED used about 50 g per day of extra-virgin olive oil — a substantial dose. Extra-virgin olive oil contains polyphenols (mainly oleuropein and hydroxytyrosol) alongside oleic acid; the polyphenols plausibly drive most of the anti-inflammatory and endothelial benefit. A re-analysis of the PREDIMED cohort tracking total polyphenol intake found the highest polyphenol fifth had 37% lower all-cause mortality versus the lowest (hazard ratio 0.63, 95% CI 0.41–0.97), with the strongest signal from stilbene and lignan intake — though the dose-response trend across the fifths was not significant (P=0.12), the whole analysis rests on 327 deaths, and it compares people within the trial by what they reported eating rather than by what they were randomised to.^[13] One detail cuts against reading this as the Mediterranean diet's mechanism: the association held in the low-fat control arm (0.48, 0.23–0.98) and in neither Mediterranean arm. A 2024 review pooling 7 cohort studies across nearly 180,000 adults found roughly 7% lower all-cause mortality at higher total polyphenol intake.^[14]

2. Plant protein replacing red meat. The pattern's protein architecture — legumes, nuts, fish — supplies the amino acids needed to maintain muscle without the strong growth signal — through insulin-like growth factor 1 (IGF-1) and the growth-driving pathway mTOR — that comes with high red and processed meat intake. See Protein for the protein side of this trade-off.

3. Microbiome and "inflammaging." Pilot studies show a switch from a Western to a Mediterranean pattern measurably alters gut microbial composition and tryptophan metabolism within four days. Fiber from unrefined grains and legumes plus polyphenols from olive oil and vegetables feed beneficial taxa, reinforce the gut barrier, and dampen the chronic low-grade inflammation that drives cardiovascular and neurodegenerative disease.

The MIND diet: optimized for brain health

The MIND diet (Mediterranean-DASH Intervention for Neurodegenerative Delay) combines elements of Mediterranean and DASH with specific emphasis on foods linked to cognitive outcomes.

MIND-specific recommendations

These are the servings that earn a full point on a 0–15 score, with half-points in between — targets rather than pass/fail rules.^[15]

Increase:

Green leafy vegetables (6+ servings/week — uniquely emphasized)
Other vegetables (1+ serving/day)
Berries (2+ servings/week — uniquely emphasized over other fruits)
Nuts (5+ servings/week)
Olive oil (primary fat)
Whole grains (3+ servings/day)
Fish (1+ serving/week)
Beans (3+ servings/week)
Poultry (2+ servings/week)
Wine (1 glass/day — outdated guidance per modern alcohol evidence)

Limit:

Red meat (<4 servings/week)
Butter / stick margarine (<1 tablespoon/day)
Cheese (<1 serving/week)
Pastries / sweets (<5 servings/week)
Fried / fast food (<1 serving/week)

Evidence

A 2015 cohort study in Alzheimer's & Dementia (923 older adults, followed 4.5 years) found the highest third for MIND adherence had ~53% lower Alzheimer's risk; the middle third about 35% lower.^[16] That is the largest estimate in the literature and it comes from the smallest study. Pooling 11 cohorts, 224,049 people and 5,279 dementia cases puts the top-versus-bottom effect at about 17% (hazard ratio 0.83, 95% CI 0.76–0.90) — a third of the headline.^[17]
Smaller benefit is observed even with moderate adherence.
The 2023 MIND-USDA trial randomised 604 cognitively unimpaired older adults — selected for family history of dementia, excess weight and a poor baseline diet rather than for any cognitive measure — to the MIND diet or their usual diet, with both arms asked to cut about 250 kcal a day. Over three years both improved and the gap between them was 0.035 standard units (95% CI −0.022 to 0.092, P=0.23) — well inside the range expected from chance. Brain-imaging endpoints were null too, and the two arms lost near-identical weight.^[18]

The 2023 RCT moderates the original observational claim. Plausible interpretation: the MIND pattern is not uniquely transformative beyond a generally healthy Mediterranean-style diet, but the broader pattern still matters. The Mediterranean figures in the 2025 meta-analysis — 18% for cognitive impairment, 11% for dementia, 30% for Alzheimer's, three outcomes in three study pools — line up with this: whatever lift there is comes from the shared core, not from MIND's specific tweaks.

DASH: blood pressure-focused

Dietary Approaches to Stop Hypertension — RCT-validated specifically for blood pressure reduction.

Components

High in fruits, vegetables, whole grains, low-fat dairy
Moderate fish, poultry, beans, nuts
Low in saturated fat, red meat, sugar-sweetened beverages
Sodium-restricted version is the most effective for BP

Evidence

The original DASH trial fed 459 people for eight weeks and lowered systolic blood pressure by about 3.5 mmHg in the 326 participants without hypertension and about 11 mmHg in the 133 who had it; the widely quoted ~6 mmHg is the whole-trial average, not the effect in people with normal pressure.^[19] Across the 30 randomised trials pooled since (5,545 participants) the typical effect is smaller still — about 3 mmHg systolic and 2.5 mmHg diastolic — and, unlike in the original trial, it did not differ by whether participants had hypertension.^[20] Moderate.
The DASH-Sodium trial showed cutting sodium buys more on top of DASH, though the two levers overlap rather than stack: high-to-low sodium was worth about 6.7 mmHg systolic on the control diet but about 3.0 mmHg on top of DASH.^[21]
Pooling 17 cohort estimates from 13 publications, each 5-point rise on a standardised DASH adherence score tracked with about 5% lower all-cause mortality (hazard ratio 0.95, 95% CI 0.94–0.96, from 14 of those estimates) and about 4% lower cardiovascular mortality (0.96, 0.95–0.98) — small per-step effects that compound across the range of the score, but with very high variation between studies, and observational throughout, so read them as "people who eat this way die later" rather than as a demonstrated effect of the diet.^[22] Moderate.

Overlap with Mediterranean

DASH and Mediterranean are largely overlapping food patterns. The main differences: Mediterranean uses olive oil more heavily and includes fish more centrally; DASH emphasizes low-fat dairy and explicit sodium restriction.

Nordic diet: the regional sister pattern

The Nordic diet (Denmark, Norway, Finland, Sweden) is structurally a Mediterranean variant adapted to colder regions: rapeseed/canola oil instead of olive oil (high in alpha-linolenic acid), rye/barley/oats instead of wheat, and locally available berries (lingonberries, bilberries) supplying anthocyanins. Pooling 15 randomised trials, Nordic-diet participants improved on blood pressure, fasting glucose, insulin and blood lipids overall — though not on high-density lipoprotein cholesterol or apolipoprotein A1 specifically, and not on inflammation, where C-reactive protein came back null.^[23] The direct comparison with the Mediterranean pattern favours the Mediterranean one: scored in the same 38,428 Swedish women over a median 17 years, high adherence to a modified Mediterranean score carried 24% lower all-cause mortality (hazard ratio 0.76, 95% CI 0.70–0.81) against 11% for the Healthy Nordic Food Index (0.89, 0.83–0.96), and the Nordic score added nothing once the Mediterranean one was accounted for.^[24] Moderate for the lipid and blood-pressure effects; the pattern's evidence base is real but shallower than the Mediterranean one. Useful to know if olives and EVOO are not local, regional staples — the pattern is what matters, not the geography.

Blue Zones: the behavioral commonalities

The five identified Blue Zones (Sardinia, Okinawa, Nicoya, Ikaria, Loma Linda) share dietary patterns that align with Mediterranean/MIND principles, plus some additional behavioral elements:

Dietary commonalities:

Plant-dominant eating — the familiar "95% plant" figure is Buettner's popular framing rather than a measured cross-zone value, and the one hard number behind it is the reconstructed 1949 Okinawan food record, where animal foods supplied only about 1–2% of calories^[25]
Beans/legumes as a daily protein staple (Sardinian fava beans, Okinawan soy/tofu, Costa Rican black beans)
Starchy staples — the traditional Okinawan diet was built on the sweet potato, over half of daily calories — plus whole grains (whole-grain bread, corn tortillas)
Modest meat consumption — Buettner's team put it at about five times a month in 85–115 g portions, a field estimate rather than a measured intake
Fermented foods (Okinawan miso and tofu, Sardinian sourdough and aged cheese)
Regular moderate alcohol in four of the five zones — wine in the Mediterranean ones, though this is the commonality that has aged worst; see Alcohol
"Hara Hachi Bu" — Okinawan principle of eating to 80% fullness
The smallest meal in the late afternoon or early evening, with nothing after it

Important caveat: Some demographic claims about Blue Zones have been challenged — clerical errors, lack of birth records, and possible welfare fraud may inflate centenarian counts in some regions. That critique is serious and partly corroborated, but as of mid-2026 it remains an unreviewed preprint, and demographers who validated the Sardinian records have answered several of its points directly — including that the critique concerns supercentenarians aged 110 and over, while Blue Zones demography rests on the proportion of a population reaching 90.^[26] ^[27] Blue Zones weighs both sides. The dietary patterns documented in those populations remain valid even if absolute longevity claims are softer.

See Blue Zones for fuller treatment.

"Plant-based" is not enough — quality matters

A common misread of the longevity diet evidence is that any plant-forward eating is automatically protective. The data say otherwise. The Plant-Based Diet Index splits plant-forward diets into a healthful version (hPDI: whole grains, legumes, fruit, vegetables, nuts) and an unhealthful version (uPDI: refined grains, fruit juices, sugar-sweetened beverages, sweets, processed plant foods). A 2025 dose-response meta-analysis of prospective cohorts found high hPDI adherence is associated with significantly lower all-cause mortality, while high uPDI adherence is associated with higher mortality — even though both diets are technically "plant-based."^[28]

Practical implication: removing meat without replacing it with whole, unprocessed plants — Coke and pasta-and-cookies vegetarianism — is not a longevity strategy. What the quality index does not explain is the ranking between the good patterns: when all eight were scored in the same 105,015 people, the healthful plant-based index came last, not first.^[29] Plant quality is a floor to clear, not the thing that separates the winners.

Comparing the patterns head-to-head

The most directly on-topic recent evidence compares the major patterns against each other for healthy aging, not just mortality. Pooling the Nurses' Health Study and Health Professionals Follow-Up Study (n=105,015; up to 30 years of follow-up), a 2025 analysis defined healthy aging as reaching 70 free of 11 chronic diseases with intact cognitive, physical, and mental function — achieved by 9.3% of participants. It scored eight patterns (AHEI, alternate Mediterranean, DASH, MIND, healthful plant-based index, Planetary Health Diet, and two insulinaemic/inflammatory indices). Moderate. Every pattern helped: odds ratios for the top versus bottom quintile ran from 1.45 (healthful plant-based) to 1.86 (Alternative Healthy Eating Index, the top performer — 86% greater odds of healthy aging at 70). Higher ultra-processed-food intake went the other way, with 32% lower odds. The cohort was predominantly white health professionals, limiting generalisability.^[30]

This is the central message in hard numbers: every pattern helped, and the top performers share the same foods. The spread is real, though — on the authors' own pairwise tests the Alternative Healthy Eating Index beat MIND, the healthful plant-based index and the inflammatory index, and the best and worst scores' intervals do not overlap. Getting the shared core right buys most of the benefit; the label is not quite arbitrary. Three things keep this modest: the eight scores are heavily correlated with one another, the pairwise tests were not adjusted for multiple comparisons, and even at high adherence the absolute chance of aging healthily runs 8.4–12.4% against a 9.3% base rate.

The Planetary Health Diet (EAT-Lancet)

The EAT-Lancet "planetary health" pattern — plant-forward, with environmental sustainability built in — has accumulating cohort support. A 2026 systematic review (227 studies; 79 meta-analysed) found higher adherence associated with roughly 20% lower all-cause mortality (pooled HR 0.80, 95% CI 0.76–0.85; n≈1.06 million) and about 17% lower cardiovascular disease (HR 0.83), no association with dementia, and none with overall cancer incidence — though cancer mortality was about 10% lower. Weak for mortality — the review's own certainty grading for both the mortality and the cardiovascular estimates is very low, with more than 75% of the variation between studies unexplained.^[31] No randomised trial of this pattern exists, so causal claims rest entirely on observational cohorts. Standard critiques apply: it may need attention to iron and vitamin B12 adequacy, affordability and cultural inclusivity are real concerns, and the EAT-Lancet Commission's own framing — its 2025 second edition claims 27% lower risk of premature death and around 15 million premature deaths averted a year — is advocacy as much as epidemiology.^[32]

Macronutrient quantity: the carbohydrate U-shape

Carbohydrate amount relates to mortality in a U-shape. In the ARIC cohort (n=15,428, 25-year follow-up) plus a meta-analysis of 432,179 people, lowest mortality fell at 50–55% of energy from carbohydrate; both low (<40%) and high (>70%) intakes raised risk (pooled HR ~1.20 for low-carb, ~1.23 for high-carb). Moderate. The decisive factor was the substitute: mortality rose when carbohydrate was replaced by animal-derived fat and protein, but fell when replaced by plant-derived sources.^[33] Low-carb meta-analyses echo this: a 2023 dose-response analysis found moderate low-carb scores most favourable for cardiovascular mortality, with very-low-carb/ketogenic patterns less so, and ketogenic diets raise LDL ("bad") cholesterol in trials without demonstrated long-term mortality benefit.^[34] This single substitution principle unifies the low-carb and plant-based stories: it is the source, not the macronutrient label, that tracks longevity.

It's never too late: improving diet quality later still helps

You do not need to have eaten well your whole life. Tracking ~74,000 adults over 12 years, the largest improvers in diet quality versus stable eaters had 9–16% lower all-cause mortality (AHEI HR 0.91, 95% CI 0.85–0.97; alternate Mediterranean 0.84, 0.78–0.91; DASH 0.89, 0.84–0.95) — the paper's own headline "8–17%" describes a 20-percentile increase in score rather than these three contrasts. Worsening diet quality raised mortality. Moderate.^[35] This is the hard-number version of the page's "you don't need to be perfect" message — mid-life course corrections register.

Vegetarian and vegan cohorts

Removing meat is not automatically protective; it depends on what replaces it. The Adventist Health Study-2 originally found all vegetarians combined had 12% lower all-cause mortality than nonvegetarians (hazard ratio 0.88), with pesco-vegetarians faring best.^[36] The same cohort's 2024 extension — 88,400 people, 12,515 deaths, median 11 years — shows the advantage is age-dependent and sex-dependent: 11% lower at age 65 (0.89, 95% CI 0.83–0.95) but gone by 85 (0.98, 0.91–1.04), and driven by men (0.81, 0.73–0.90) rather than women (0.94, 0.86–1.02). It also found signals in the other direction in late old age — stroke 17% higher (1.17, 1.02–1.33) and dementia 13% higher (1.13, 1.00–1.27) — which the authors state plainly.^[37] Weak / preliminary.

A pooled analysis of two UK cohorts — EPIC-Oxford and the Oxford Vegetarian Study, 60,310 people and 5,294 deaths — found essentially no all-cause mortality difference between vegetarians and regular meat eaters (hazard ratio 1.02, 95% CI 0.94–1.10).^[38] The honest reading is a trade-off rather than a wash. The label does buy something: pooled across prospective studies, vegetarians had 13% lower total cancer incidence and vegans 23% lower (0.87, 95% CI 0.84–0.91 and 0.77, 0.70–0.85).^[39] It also costs something: over 18 years EPIC-Oxford vegetarians had 20% more strokes (1.20, 1.02–1.40) — about 3 extra cases per 1,000 people over 10 years, mostly haemorrhagic — alongside 22% less ischaemic heart disease.^[40] So plant-food quality is the lever for how well a plant-forward diet works; it is not the whole story of what dropping meat does.

Biological aging: the strongest causal signal is caloric restriction

The cleanest randomised human evidence in this field comes from caloric restriction — though it lands on a biological-age marker, not on how long anyone lived. The CALERIE randomised trial (220 healthy non-obese adults assigned to a 25% calorie-restriction target or unrestricted eating for two years) slowed the pace of biological aging measured by the DunedinPACE DNA-methylation clock by roughly 2–3%. Two deflators belong with that every time it is quoted: participants achieved a mean of about 12% restriction, not the 25% prescribed, and the methylation analysis was performed post hoc on the 197 participants with usable samples rather than as the trial's designed test. Moderate (biomarker endpoint, not mortality). Investigators likened the magnitude to a smoking-cessation effect and extrapolated it to ~10–15% lower mortality risk — but that mortality figure is an extrapolation from other DunedinPACE studies, not measured in CALERIE, and the trial did not move the PhenoAge or GrimAge clocks.^[41] Mediterranean-adherence sub-studies link the pattern to telomere maintenance, but that human telomere data is mixed and largely cross-sectional — suggestive, not definitive. See epigenetic alterations for the clocks themselves.

The analysis linking eating windows under 8 hours to cardiovascular death drew heavy media coverage as a 2024 conference abstract, and has since been published in full. In a US nutrition survey followed for a median of 8.1 years, people reporting an eating window under 8 hours had more than double the cardiovascular mortality of those eating across 12–14 hours (hazard ratio 2.35, 95% CI 1.39–3.98); the all-cause association was borderline and did not survive most sensitivity analyses, and the cancer association was absent.^[42] Caution — the estimate rests on 383 people, about 2% of the sample, contributing 33 cardiovascular deaths, who were more often smokers, food-insecure and on low incomes, and who were never shown to be fasting deliberately; the authors themselves ask whether the signal is the short window or the confounding around it. The full weighing is under Fasting.

What all four patterns have in common

The 80/20 pattern: get the food categories roughly right, and the specifics matter less.

Default toward:

Vegetables and fruit (multiple colors, including leafy greens daily)
Legumes (3+ times/week)
Whole grains over refined
Fish over red meat
Nuts and seeds
Plant oils over butter — olive oil first for the Mediterranean pattern's sake, but rapeseed/canola and soybean oil carry the same advantage over butter (per 5 g/day, about 15% lower total mortality for canola, 8% for olive, 6% for soybean), and the "seed oils are toxic" framing does not survive the evidence^[43] — see Dietary fats
Water and unsweetened beverages

Limit:

Ultra-processed foods — the most consistently replicated dose-response signal in modern nutrition, though not a high-certainty one: highest versus lowest intake carries about 15% higher all-cause mortality (hazard ratio 1.15, 95% CI 1.09–1.22), rising somewhere between about 3% and 10% per 10% increment of ultra-processed food in the diet — the two 2025 dose-response meta-analyses disagree threefold and neither has settled it.^[44] ^[45] Moderate (the 2025 US Dietary Guidelines committee rated the evidence only "limited" because NOVA-based definitions are inconsistent). See Ultra-processed food.
Sugar-sweetened beverages
Processed meats (charcuterie, bacon, sausages, hot dogs)
Excessive red meat
Refined grains and added sugars
Excessive sodium (especially from processed foods)

Practical implementation

A simple weekly template

Daily: salad/leafy greens, fruit (especially berries), olive oil, nuts, water
Most days: whole grains, legumes, vegetables in volume
2–3×/week: fish (especially fatty)
Occasional: poultry, eggs
1×/week or less: red meat
Rarely: processed meat, fried fast food, sweets

Cooking principles

Cook from whole ingredients when possible
Use olive oil generously — its place here comes from the Mediterranean pattern and its polyphenols rather than from outranking other plant oils, which the dedicated cohort analysis finds it does not^[46]
Make legumes the default protein for several meals/week
Build meals around vegetables, not as garnish
Eat the rainbow — diversity of plant colors = diversity of polyphenols

Realistic expectations

You don't need to be 100% Mediterranean. Cohort data show even moderate adherence captures meaningful benefit. The aim is the pattern dominating your eating, not perfect compliance.

If you are over 70

Two things change. The endpoint that matters shifts from mortality toward staying independent, and here the pattern still earns its place: pooling nine cohorts and 94,072 older adults, each one-point rise in Mediterranean adherence tracked with about 5% lower risk of becoming frail (odds ratio 0.95, 95% CI 0.93–0.97, moderate certainty) — though the same review found no effect on the onset of disability.^[47] The second change is protein. Anabolic resistance means older muscle needs a larger amino-acid load to respond at all, so "make legumes the default protein" needs pairing with a deliberate target — see Protein, which sets 1.0–1.5 g/kg/day and 30–40 g per meal past 65. A plant-forward plate and adequate protein are compatible, but past 70 the second no longer takes care of itself.

Dietary patterns

Dietary patterns

The Mediterranean diet: the strongest evidence base

Core components

Hard outcome data

What's actually doing the work

The MIND diet: optimized for brain health

MIND-specific recommendations

Evidence

DASH: blood pressure-focused

Components

Evidence

Overlap with Mediterranean

Nordic diet: the regional sister pattern

Blue Zones: the behavioral commonalities

"Plant-based" is not enough — quality matters

Comparing the patterns head-to-head

The Planetary Health Diet (EAT-Lancet)

Macronutrient quantity: the carbohydrate U-shape

It's never too late: improving diet quality later still helps

Vegetarian and vegan cohorts

Biological aging: the strongest causal signal is caloric restriction

What all four patterns have in common

Practical implementation

A simple weekly template

Cooking principles

Realistic expectations

If you are over 70

Further reading

More in Nutrition