Key takeaways
- The three are not comparable categories: Mediterranean sets the food, keto sets a metabolic state, high-protein sets one number and leaves the rest open.
- Mediterranean has the most hard-endpoint data, and several of its mechanisms — fat quality, fibre, polyphenols, displacing processed food — run in the same direction.
- Keto's early weight change is substantially glycogen and water; lipids move unpredictably, so a panel including ApoB a few months in is not optional.
- High-protein is the only one of the three that actively defends lean tissue in a deficit, through satiety, thermic effect and substrate supply.
- Adherence explains most of the gap between controlled-feeding results and free-living results — the pattern you can still run in two years wins.
These three are usually presented as competitors, which is the first mistake. They are not the same kind of thing. Mediterranean is a food pattern — it tells you what to put on the plate. Keto is a metabolic state — it tells you what to avoid in order to force a fuel switch. High-protein is a macronutrient priority — it tells you one number to hit and leaves the rest open. You can follow two of them at once without contradiction, which is why the honest answer to "which wins" is rarely one name.
Mediterranean
Pattern: vegetables, fruit, whole grains, legumes, nuts, olive oil as the primary fat, fish, modest poultry, eggs and dairy, minimal red meat, modest red wine. High fibre, moderate protein, low refined carbohydrate.
Strongest evidence. The PREDIMED trial (n=7,447) showed roughly a 30% reduction in major cardiovascular events versus control (Estruch et al., N Engl J Med 2018). Cohort evidence points the same way, with greater adherence associated with lower all-cause mortality across large pooled populations (Sofi et al., BMJ 2008). No other dietary pattern has this much hard-endpoint data behind it.
Why it plausibly works. Several mechanisms run in the same direction at once, which is probably the point: monounsaturated fat and marine omega-3 replacing saturated fat shifts the lipid profile; high fibre feeds the microbiome and blunts post-meal glucose; polyphenols reduce oxidative and inflammatory signalling; and the pattern displaces ultra-processed food almost automatically. Trials that tried to isolate a single nutrient have generally underperformed the whole pattern.
Limitations. Protein is moderate by design, which is fine for a sedentary population and not enough for anyone training seriously or defending muscle in midlife. It is also not weight-loss optimised: the trial evidence is about cardiovascular events, not about the scale.
Keto
Pattern: very low carbohydrate (under about 30 g per day), high fat (roughly 70% of calories), moderate protein. Sustained carbohydrate restriction depletes liver glycogen, insulin falls, fat is mobilised, and the liver produces ketone bodies the brain can use in place of glucose. It was developed for epilepsy, and that is still its strongest clinical indication.
Strengths. Strong satiety in many people. Effective for type 2 diabetes management, lowering HbA1c quickly (Zhou et al., Int J Environ Res Public Health 2022). Some evidence for cognitive support in specific neurological conditions. Early weight loss can be substantial — though a meaningful share of the first fortnight is glycogen and its associated water, which is worth knowing before the first weigh-in sets expectations.
Limitations. Hard to sustain, with adherence dropping sharply past six months (Gardner et al., JAMA 2018). Lipids move unpredictably — some people improve, and a subset see ApoB rise, which is the number that matters for cardiovascular risk and the reason a lipid panel a few months in is not optional. Fibre falls unless it is deliberately engineered back in, with the usual consequences for the gut. Long-term outcome data is sparse. Performance above moderate intensity is commonly degraded, because glycolysis is the pathway that supplies hard efforts.
High-protein
Pattern: 1.0-1.2 g of protein per pound of goal weight, with carbohydrate and fat filling the remaining calories. Less a diet than a priority applied to whatever you already eat.
Why it works mechanistically. Three separate effects stack. Protein is the most satiating macronutrient per calorie (Leidy et al., Am J Clin Nutr 2015), so intake falls without deliberate restriction. It has the highest thermic effect, meaning a larger share is spent on digestion. And it supplies the amino acids that make muscle protein synthesis possible, which is what determines whether weight lost in a deficit comes off as fat or as lean tissue.
That third point is the one that separates it from the other two frameworks. Mediterranean and keto are both compatible with losing muscle alongside fat. A protein-prioritised approach combined with resistance training is the only one of the three that actively defends it. See how much protein you actually need.
Limitations. It says nothing about the rest of the plate. Without parallel attention to fibre and food quality it degenerates into protein, cheese and not much else — which hits the protein target and misses everything the Mediterranean evidence is built on.
Side-by-side
| Factor | Mediterranean | Keto | High-Protein |
|---|---|---|---|
| Cardiovascular outcomes | Strongest evidence | Mixed | Generally good |
| Weight loss (short-term) | Modest | Strong initially | Strong |
| Long-term sustainability | High | Low | High |
| Muscle preservation | Moderate | Moderate | Strongest |
| Athletic performance | Good | Mixed | Good |
| Fiber intake | High | Often low | Variable |
| Type 2 diabetes | Effective | Very effective short-term | Effective |
The variable that actually decides it
Controlled feeding trials that match calories and protein tend to find small differences between dietary patterns. Free-living trials find larger ones. The gap between those two findings is adherence, and adherence is not a character trait — it is a property of how well a pattern fits a life.
Three questions predict it better than any macronutrient argument. Can you eat this where you actually eat — canteens, other people's houses, restaurants? Does it survive a bad week? Can you still eat it in two years? The trial data reflects exactly this: the diets that lose their advantage by twelve months are usually the ones people stopped following at month five.
This is also why "which diet is best" is a poorly formed question at the individual level. The best diet in the literature and the best diet for a given person are only the same thing if that person can run it.
The hybrid that works
The framework most longevity-focused practitioners now converge on:
- Mediterranean-style food choices — vegetables, fish, olive oil, legumes, whole grains, nuts
- High-protein emphasis, 1.0+ g per pound of goal weight
- Moderate carbohydrate — not keto, not high
- Limited refined sugar and processed food
- Limited alcohol
This is not a compromise; it is the combination of the pattern with the best hard-endpoint evidence and the macronutrient with the best body-composition evidence. They do not conflict. Fish, legumes, dairy and poultry are Mediterranean staples and protein sources simultaneously — the adjustment is mostly one of portion and frequency rather than of food choice. Practical templates are in the high-protein recipe library, and the fibre gap covers the input most people are furthest from meeting.
Which when
| Goal | Best fit |
|---|---|
| Cardiovascular longevity | Mediterranean (or hybrid) |
| Maximum body composition | High-protein hybrid |
| Type 2 diabetes management | Keto or low-carb |
| Sustainable long-term | Mediterranean / hybrid |
| Athletic performance | High-protein with adequate carbs |
One addition to that table: if the main problem is post-meal glucose rather than weight, carbohydrate quality and meal composition move the needle before total carbohydrate restriction does — glucose variability and insulin sensitivity cover the levers in order.
What to expect, and when
Timelines differ enough between these that mismatched expectations cause more abandonment than any nutritional problem.
Keto moves the scale fastest and least meaningfully at first — glycogen and water dominate the first two weeks, and a period of fatigue and headache while the fuel switch completes is common. Judge it at eight to twelve weeks, and get a lipid panel including ApoB in that window rather than assuming the direction.
High-protein shows up first as reduced hunger, usually within one to two weeks, and as retained strength during a deficit over the following months. The scale moves more slowly than on keto and the mirror moves faster, because the composition of the loss is different.
Mediterranean produces the least dramatic short-term feedback of the three. Its evidence base is measured in cardiovascular events over years, and markers such as ApoB and hs-CRP are the reasonable interim readouts. If you need visible progress in three weeks to stay motivated, pair it with the protein target rather than abandoning it.
The principle: the right diet depends on the goal, and the three frameworks answer different questions. For most adults pursuing healthspan, Mediterranean food choices with a protein target on top is the answer — it takes the pattern with the strongest outcome data and adds the one thing that pattern is missing. Pure keto is for specific clinical situations, not a general upgrade.
Bottom line
Mediterranean has the strongest long-term outcome evidence. High-protein has the strongest body-composition evidence. Keto has genuine clinical applications and a weak long-term record outside them. Because they operate on different axes — food choice, metabolic state, macronutrient priority — the sensible answer for most healthspan-focused adults is a Mediterranean-style high-protein hybrid rather than allegiance to one label. Whichever you choose, adherence over years beats optimisation over weeks, and a framework you can still run in two years will outperform a better one you abandoned in month five.
Educational content, not medical advice. Laboratory interpretation and any treatment decision are made by a licensed physician after individual evaluation. Individual results vary.
