Longevity diet: what to eat to actually live longer

Apparently the easiest way to live longer, according to most of the research that’s survived peer review, is not to discover some new superfood. It’s mostly to eat vegetables, beans, olive oil, fish, and a reasonable amount of everything else — and to not eat very much ultra-processed food. Which is a deeply unsexy finding. The wellness industry has spent two decades trying to make it sound more exciting than it is.

But the evidence is quite good actually, and the gap between what the research says and what the internet says is now wide enough that it’s worth walking through properly.

Diet shows up in every major longevity dataset as one of the top three or four modifiable factors. Alongside muscle mass, cardiorespiratory fitness, and sleep, what you eat shapes biological age in ways that compound over decades. Unlike supplements or biohacks, which tend to move biomarkers by rounding errors, dietary pattern is the intervention with effect sizes you can actually see in mortality curves.

Here’s what the research actually supports, what it doesn’t, and what a longevity-oriented diet looks like on a Tuesday.

The Mediterranean diet: boring, researched, effective

The single most studied dietary pattern in longevity research is the Mediterranean diet — and when I say studied, I mean thousands of papers, multiple large randomised trials, and more than 40 years of prospective cohort data. It is not trendy. It has no hashtag. It just keeps winning.

The landmark trial is PREDIMED — Prevención con Dieta Mediterránea — a Spanish multi-centre study that enrolled 7,447 adults at high cardiovascular risk and randomised them to a Mediterranean diet supplemented with either extra-virgin olive oil or mixed nuts, versus a low-fat control diet. The results were so striking that the trial was stopped early for ethical reasons: after a median follow-up of 4.8 years, the Mediterranean groups had roughly a 30% reduction in major cardiovascular events compared to the control. Estruch and colleagues published the updated analysis in the New England Journal of Medicine in 2018. Thirty percent is not a rounding error. That’s the kind of effect size drug companies design trials around.

Cohort data reinforces the trial data. Sofi and colleagues’ 2010 meta-analysis in the American Journal of Clinical Nutrition aggregated prospective studies covering more than 2 million participants and found that each 2-point increase on a Mediterranean adherence score was associated with a 9% reduction in all-cause mortality, a 9% reduction in cardiovascular mortality, a 6% reduction in cancer mortality, and a 13% reduction in neurodegenerative disease. The direction was consistent across every population studied.

What makes the Mediterranean diet work isn’t any single food. It’s the composition:

• Vegetables at most meals — 400-500 g a day is typical in adherent populations
• Legumes (chickpeas, lentils, white beans) at 2-4 servings a week, often more
• Whole grains — bread, pasta, farro, bulgur — rather than refined
• Extra-virgin olive oil as the main cooking fat, generously used
• Fish and seafood 2-3 times a week
• Nuts and seeds as regular snacks
• Dairy in moderate amounts, often fermented (yogurt, cheese)
• Poultry and eggs a few times a week
• Red meat rarely, processed meat almost never
• Wine with meals, modest amounts, for those who drink

None of that is exotic. None of it requires expensive products. Most of it is cheaper than the standard Western plate. And yet the mortality effect is roughly what you’d expect from a solid pharmaceutical intervention, only without the side effects.

Blue Zones: the observational version

The Blue Zones — Sardinia (Italy), Okinawa (Japan), Nicoya (Costa Rica), Ikaria (Greece), and Loma Linda (California) — are five regions where people reach 100 at extraordinary rates. Dan Buettner’s team, working with demographers including Gianni Pes and Michel Poulain, spent years characterising what these populations eat. The results published in the American Journal of Lifestyle Medicine in 2016 surprised nobody who’d looked at the Mediterranean data.

The common threads across all five regions:

• 95% or more of calories from plants — vegetables, legumes, tubers, whole grains
• Beans as a dietary staple — a cup a day in most Blue Zones
• Moderate intake — they stop eating when about 80% full, a habit the Okinawans call “hara hachi bu”
• Meat sparingly — roughly 5 times per month, in small portions, often as flavouring rather than the centre of the plate
• Wine in moderation — 1-2 small glasses a day, almost always with food and friends
• Water, coffee, tea — minimal sugary drinks

The Blue Zones research is observational, so you can’t draw causal conclusions the way you can from PREDIMED. But it’s genuinely striking that five unrelated populations on three continents, with different cuisines and cultures, independently converge on a pattern that looks remarkably like what the clinical trials also support. Legumes, vegetables, olive oil, whole grains, minimal processed food, moderate overall intake. The convergence is the finding.

Protein: where longevity nutrition gets slightly argumentative

Protein is the one place where the longevity-nutrition conversation gets genuinely contested. There are two camps.

The first says protein intake should be modest, because elevated levels of IGF-1 (insulin-like growth factor 1) — which protein and especially animal protein drive upward — are associated with increased cancer risk in some cohorts. Valter Longo’s group has been the loudest voice here, arguing for lower protein intake in midlife followed by higher intake after 65.

The second says under-eating protein drives sarcopenia, which is itself one of the strongest predictors of mortality in older adults. Stuart Phillips’ group at McMaster has published extensively arguing that the 0.8 g/kg/day RDA is outdated and that 1.2-1.6 g/kg is closer to optimal for healthy adults.

The honest answer is that both are right about different things. For most adults under 65 who are actively training, 1.2-1.6 g/kg/day is supported by the bulk of the evidence. For adults over 65, the need goes up, not down, because muscle protein synthesis becomes less efficient — Bauer and colleagues’ 2013 consensus paper in JAMDA recommended 1.0-1.2 g/kg/day as a floor for healthy older adults, with higher needs during illness or training. The midlife question is the only genuine live debate, and even there, the effect sizes are smaller than the muscle-preservation effect of eating enough.

The more important finding, which both camps tend to agree on, is that the source of protein matters. Song and colleagues published a large analysis in JAMA Internal Medicine in 2016 using the Nurses’ Health Study and Health Professionals Follow-up Study — over 130,000 adults with up to 32 years of follow-up. They found that replacing 3% of energy from animal protein with plant protein was associated with a 10% reduction in all-cause mortality, with the strongest effect from replacing processed red meat. Plant protein intake was associated with lower mortality regardless of baseline diet.

Which fits with the Mediterranean and Blue Zones patterns, where most protein comes from legumes, fish, nuts, and modest amounts of dairy rather than from daily beef. More on muscle-specific protein targets in the muscle mass and longevity post — the nutrition and training sides have to rhyme.

Ultra-processed food: the thing actually worth worrying about

If the last section sounded argumentative, this one is the opposite. The evidence on ultra-processed food is grim and extraordinarily consistent.

The NOVA classification system, developed by Carlos Monteiro’s group in Brazil, divides foods into four categories: unprocessed, minimally processed, processed, and ultra-processed. Ultra-processed foods are industrial formulations typically containing ingredients you wouldn’t find in a home kitchen — modified starches, emulsifiers, flavour enhancers, artificial sweeteners. Most packaged snacks, breakfast cereals, soft drinks, and ready meals fall into this category.

Srour and colleagues published a 2019 analysis in BMJ following 105,000 French adults over 5 years. Each 10% increase in the proportion of ultra-processed food in the diet was associated with a 14% increase in all-cause mortality. A 2019 JAMA Internal Medicine analysis by Rico-Campà and colleagues of 19,899 Spanish adults found similar: those in the highest quartile of ultra-processed food intake had a 62% higher risk of all-cause mortality compared to the lowest quartile.

A randomised controlled trial by Hall and colleagues at the NIH, published in Cell Metabolism in 2019, is particularly clean. They put 20 adults on two diets for two weeks each — matched for calories, macros, sugar, fat, fibre — with the only difference being whether the food was ultra-processed or not. On the ultra-processed diet, participants spontaneously ate 500 extra calories per day and gained 0.9 kg on average. Even with matched nutrients, the processing itself drove overconsumption.

The honest practical translation: the single highest-leverage dietary change most people can make for longevity is reducing ultra-processed food. Not eliminating it — that’s neither realistic nor necessary — but shifting the balance toward food you’d recognise as food.

Fasting and caloric restriction: the genuinely interesting edge

Time-restricted eating — compressing your daily eating window to 8-12 hours — has exploded in popularity, and the underlying research is more interesting than the hype suggests.

Caloric restriction extends lifespan in basically every species tested, from yeast to monkeys. The CALERIE trial, published in The Lancet Diabetes & Endocrinology in 2019, was the longest and most rigorous human study: 218 non-obese adults randomised to either maintain their habitual diet or reduce calories by roughly 15% for two years. The restricted group showed improvements in cardiometabolic risk markers, insulin sensitivity, inflammation, and a slowed pace of biological ageing on epigenetic clocks. Whether this translates to actual lifespan extension in humans is still unknown — nobody’s going to run a 50-year trial.

De Cabo and Mattson’s 2019 review in the New England Journal of Medicine is the cleanest summary of the intermittent fasting evidence. Their conclusion, roughly: time-restricted eating improves insulin sensitivity, reduces inflammation, lowers cardiovascular risk factors, and improves cognitive markers in humans. Whether it extends lifespan in humans specifically is unproven, but the mechanistic story is coherent enough that it’s probably worth integrating for most healthy adults.

The practical version is unglamorous. Eating within a 10-12 hour window — say, 8 AM to 7 PM — captures most of the benefit without requiring you to fast aggressively. The 16:8 protocols work for some people and not others. Longer fasts (24+ hours) remain more experimental and probably aren’t necessary for the average person looking to age well.

Common misconceptions

A few things that come up constantly:

• “Carbs are bad for longevity.” They’re not, on average. The populations with the longest lifespans in recorded history — Okinawans, Sardinians, Ikarians — eat substantial amounts of carbohydrate, almost entirely from whole grains, tubers, and legumes. The problem isn’t carbohydrate as a macronutrient; it’s refined carbohydrate in ultra-processed form.
• “Fat is bad for longevity.” Also not. The Mediterranean diet is 35-40% fat, mostly from olive oil, nuts, and fish. Total fat intake doesn’t predict mortality in recent meta-analyses; the type of fat does. Polyunsaturated and monounsaturated fats reduce cardiovascular risk; trans fats raise it sharply.
• “Gluten / dairy / nightshades are silently killing you.” For the small percentage of people with actual coeliac disease or a medically confirmed intolerance, yes. For everyone else, the elimination-diet industry is working from scant evidence. None of the Blue Zones populations avoid these foods, and whole grains and fermented dairy consistently appear on the beneficial side of the ledger.
• “Supplements can replace a good diet.” They can’t. Multivitamin trials have been conducted at scale — VITAL, the Physicians’ Health Study II — and the mortality effects are essentially zero in well-nourished adults. Supplements fill specific deficiencies (vitamin D in northern latitudes, B12 in vegans) but do not replicate the effects of a good dietary pattern.
• “I need to eliminate sugar entirely.” The goalposts here are unhelpful. Added sugar should be limited — 25-50 g a day is a reasonable ceiling — but the whole-fruit sugar in a bowl of berries is not the problem. Fructose from soft drinks and the ultra-processed food supply is.

The practical protocol

If I were handing someone a one-page summary of this entire post, it would read like this:

1. Build most meals around plants. Vegetables, legumes, whole grains, nuts. Aim for half the plate, most meals.
2. Use olive oil generously. It’s the fat with the strongest longevity evidence in humans. Extra virgin, cold-pressed where you can.
3. Eat fish 2-3 times a week. Fatty fish especially — salmon, sardines, mackerel — for the omega-3s.
4. Cap red meat at 1-2 servings a week. Cap processed meat near zero. The evidence on processed meat is unambiguous and the economic cost of moderation is trivial.
5. Hit your protein target. 1.2-1.6 g/kg/day for most adults; 1.2-2.0 g/kg for adults over 65 who train. Distribute across 3-4 meals. See the muscle mass post for why this matters.
6. Limit ultra-processed food. This is the single highest-leverage change for most people. Aim for under 20% of calories; most adults currently live well above 50%.
7. Compress your eating window to 10-12 hours if it fits your life. Probably 8 AM to 7 PM or similar. Don’t eat right before sleep.
8. Keep alcohol modest. One drink a day for women, two for men, is the ceiling in most longevity guidance. The Blue Zones drink; they don’t drink much.
9. Drink water, tea, coffee. The longest-lived populations don’t drink sugar. Coffee is on the net-positive side in most cohort data; moderate intake is fine.
10. Don’t stress about the edges. The difference between a 90th-percentile longevity diet and a 99th-percentile one is smaller than the difference between a 50th-percentile diet and a 90th-percentile one. Consistency at “good enough” beats perfection you can’t sustain.

That’s genuinely most of the value. The remaining 10% is detail that only matters after the first 90% is in place.

The tracking bit

Food tracking is one of those things that’s useful for about three weeks and insufferable after six. The honest version: track for long enough to understand what your protein intake actually is, what your rough calorie ballpark looks like, and where your ultra-processed food is sneaking in. Then stop tracking and just cook.

Sarvita lets you log meals with a photo — you point your phone at the plate, it estimates protein, calories, and composition, and folds it into the picture alongside biological age, VO2 Max, and muscle trends. The point isn’t to turn eating into a spreadsheet. It’s to make sure your actual dietary pattern is close to the one you think you have. For most of us, the gap between the two is larger than we’d like to admit.

Apparently I’ve been eating significantly more olive oil than my Victorian-orphan-protein-intake self used to. Fair enough — the evidence says that’s the right direction. The thing I find quietly reassuring about longevity nutrition is that it doesn’t require perfection. It just requires making the reasonable choice about 80% of the time and not being too precious about the rest.

Anyway. Link’s there if you’d like Sarvita to track nutrition alongside the rest of it. No pressure — and the app doesn’t judge you for the pizza, which is my favourite kind of gamification.