Ask how much protein to eat and you are really asking four questions at once: how much, from which foods, at what times of day, and whether any of it applies to you. Each has a different answer, and each rests on its own evidence.
If you lift weights, protein earns a place. Reaching a modest daily target lets the training build a little more muscle than it otherwise would — and that target is a floor to reach, not a ceiling to fear. Eating past it is nearly harmless for a healthy person and adds almost nothing; without the training, extra protein on its own builds little. For how long you live, the amount barely moves the needle. What does, modestly, is the source: shifting the protein on your plate from animal toward plant tracks slightly lower mortality, though only in observational data.
Who you are changes the answer most of all. An older adult who already resistance-trains gains little from piling on protein or supplements beyond the training itself — the one thin exception a weak grip-strength signal in diagnosed muscle-wasting — while an older adult eating too little should still raise and spread out their protein. Underneath all of it sits one caveat: almost every number here is measured on muscle size and strength, not on whether people live longer or better, and the kidney scare that makes some people ration protein is unfounded unless the kidneys are already diseased.
Muscle: a floor worth reaching, a ceiling not worth chasing
Start with the muscle question, because it is the one most people mean. The lever that builds muscle is the training, not the protein. Pooling 49 resistance-training trials, Morton found that adding protein on top of the lifting bought a further 0.30 kg of lean mass and 2.49 kg on a one-rep-max over weeks of training (Morton et al., 2017) — real, but small increments on what the training already delivered, and it moved neither total body weight nor maximal voluntary contraction at all (Morton et al., 2017). Protein does not build muscle; it lets the training build slightly more.
How much protein feeds that adjunct? Morton’s dose-response points at roughly 1.6 g per kilogram of body weight per day, above which «Protein supplementation beyond total protein intakes of 1.62 g/kg/day resulted in no further RET-induced gains in FFM» (Morton et al., 2017). Two things about that number matter more than the number itself.
First, the denominator is total body weight as the scale reads it — grams of protein per kilogram of whole-body mass, not per kilogram of lean mass. Morton regressed the change in fat-free mass against baseline intake in g/kg of total body mass; fat-free and fat mass are the outcomes being predicted, not the yardstick the dose is measured against (Morton et al., 2017). In a lean person, total and lean weight sit within about 15-20% of each other, so ~1.6 g/kg of total weight is usable as-is; in someone with obesity the two diverge roughly two-fold, and the gram target is undefined until you say which kilogram you mean.
Second, the knee is soft. The break-point of 1.62 carries a wide confidence interval of 1.03-2.20 and was not statistically significant (p=0.079) (Morton et al., 2017) — it is essentially the midpoint of that interval, not a demonstrated threshold. On Morton’s own baseline-intake axis a straight line fit the data significantly better than a knee, and an independent Japanese meta-analysis (Tagawa, 138 trials, no shared authors) placed its diminishing-returns inflection lower, at 1.3 g/kg, on a curve that kept rising out to 3.5 g/kg and never flattened (Tagawa et al., 2020). Different populations, different estimators, no plateau anyone could pin down: the honest object is a wide region, not a point -> The Underivable Optimum.
So treat ~1.6 g/kg as a floor to reach, not a ceiling to fear. The failure that costs you muscle is undershooting; overshooting merely wastes a little protein, and for a healthy person, eating well past the target does not harm the kidneys (see the strata below). Most people who already resistance-train sit near the 1.4 g/kg that Morton’s trial participants averaged — about 75% above the RDA — so the practical move is topping up into the region, not loading far beyond it. (Morton et al., 2017)
Where the person leaves Morton’s lean, weight-stable, trained sample, the target is not simply lower — it is unknown. People with obesity, older adults, and anyone in a caloric deficit sit off the evidence’s support: Morton reported no obese stratum and no baseline body-fat, so the right move is to name the denominator and treat the target as a genuine gap, not to scale a lean number downward. A tempting shortcut — dose on lean mass instead — does not even settle its own direction: heavier people often carry more absolute lean mass, which pushes the gram target up, while anabolic resistance may raise the per-kilogram requirement, which pushes it the other way. Sign-uncertain, so hold it as mechanism, not a finding.
The 1.6 that matters for muscle is not the only number people quote — and the others answer different questions.
The famous numbers each answer a different question
Three protein numbers circulate, all in the same units — grams per kilogram of body weight per day — which is exactly why they get confused. They are not competing estimates of one right answer. They answer different questions.
The RDA of 0.8 g/kg is the lowest of the three and the most misread. It is a deficiency floor — the single figure every guideline sets for all adults, derived from nitrogen-balance studies — and PROT-AGE judges it «too low for older people» (Bauer et al., 2013). It was never meant as a target to aim for, and deploying it as one — the way a food label’s % Daily Value invites — undershoots anyone trying to build or hold muscle. In one 14-week study, older adults eating exactly the RDA «resulted in the loss of mid-thigh muscle area» even as their whole-body composition looked stable (Bauer et al., 2013).
The gym maxim a gram per pound lands near 2.2 g/kg once converted — at or above the top of the muscle-building region, past the point where more protein reliably adds lean mass. It is not so much wrong as redundant: it buys the upper-end margin discussed above and little beyond it.
Between them sits ~1.6 g/kg, the region where resistance training’s protein benefit flattens. Line the three up and the pattern is plain: 0.8 answers how little keeps me out of deficiency, 1.6 answers how much helps me build muscle while I train, and 2.2 is a rule of thumb sitting at the ceiling of that second question. Same unit, three different constructs — preventing deficiency versus building muscle — not three tries at one number. And every one of them is grams per kilogram of total body weight, so the denominator has to be named each time or the figures stop meaning anything.
All three numbers are about amount; for how long you live, the evidence points at source instead.
For how long you live, the source matters and the amount barely does
All three famous numbers ask how much protein. For mortality, the evidence points somewhere else entirely — at what kind. Naghshi’s meta-analysis pooled 32 prospective cohorts and found that raising total protein tracks only a weak drop in all-cause death (RR «0.94 … 0.89 to 0.99»), with no signal at all on cardiovascular or cancer mortality (Naghshi et al., 2020). When the total splits by source, the picture sharpens: animal protein is flatly null on every outcome (all-cause «1.00 … 0.94 to 1.05»), and plant protein carries the whole signal — all-cause «0.92 … 0.87 to 0.97», CVD «0.88 … 0.80 to 0.96» (Naghshi et al., 2020). The one significant dose-response is small: «an additional 3% of energy from plant proteins a day is associated with a 5% lower risk of death from all causes» (Naghshi et al., 2020).
So the lever here is not a target to hit — it is a swap. Shift protein-bearing foods from animal toward plant (legumes, nuts, whole grains, soy); the daily gram count does little on its own. This is a substitution decision, not a dose decision.
Two cautions keep the swap honest. First, the animal-protein null is a fact about the nutrient, not a green light for the foods -> Is the Food Category Doing Any Work. The animal bucket pools red meat with fish, poultry, eggs and dairy, and Budhathoki’s Japanese cohort — where fish supplies most animal protein — shows why the aggregate reads null: protective fish and harmful red and processed meat cancel inside one label, a discrepancy the authors trace to «a difference in the main dietary source of animal protein» (Budhathoki et al., 2019). Read animal protein is null as name the food — not as red meat is fine.
Second, hold the whole finding loosely. Every estimate is observational, and Naghshi’s own leading caveat
is confounding by diet pattern and social class. Plant-protein foods are also the fibre-bearing foods the
wiki already credits with lower mortality, so plant protein may be the fibre and pulse lever wearing
another name — a gap this evidence cannot close -> Dietary Fibre and Health. And no second, genuinely
independent cohort backs it up: Budhathoki sits inside Naghshi’s 32-cohort pool, and Seidelmann’s
carbohydrate-substitution study shares Willett’s authorship and the same NHS/HPFS cohorts. Agreement
across shared data is not independent replication, so the page stays confidence: low — no lift, no
[E-independent] claim.
Amount and source together set the daily total. Two second-order dials — when you eat protein and what quality it is — matter mainly at the margins, and mainly for older adults.
Spread protein across meals, and let quality matter only until you hit the target
Timing earns its place through one mechanism: Anabolic Resistance. An older muscle needs a larger per-meal stimulus to switch on the same building response a younger one reaches easily. Moore’s tracer work puts the muscle-protein-synthesis plateau near 0.40 g/kg of body mass per meal in older men, ~68% above younger men (Moore et al., 2014), and Katsanos found that a 2.8 g leucine bolus restored a synthesis response that a 1.7 g one did not (Katsanos et al., 2006) — which the wiki carries as a practical ~25-30 g of protein (about 2.5-2.8 g of leucine) per meal to clear the raised threshold .
So spread protein across the day’s meals rather than loading it into dinner: a meal that falls short of the threshold is a dose partly wasted, so how you distribute the protein matters as much as the daily total.
Quality is the second dial, and DIAAS is how it is scored — a food’s limiting amino acid times how much is digested. FAO’s own worked numbers rank plant below animal: wheat scores 40, peas 64, whole milk powder 122 (Food and Agriculture Organization of the United Nations, 2013). So hitting a protein target from low-DIAAS plant sources takes more grams, or deliberate complementation (cereal plus legume covers each other’s limiting amino acid) -> Protein Quality and the DIAAS Score.
But the quality gap is a surrogate advantage that shrinks as you approach the target. DIAAS predicts how well a protein could meet amino-acid demand; it moves no muscle or strength outcome the wiki holds evidence for once total daily protein is matched and above adequacy. Whey’s edge is real but acute and per-gram — largest in a single dose, not across a diet. For someone hitting the amount from a mixed diet with any animal protein or deliberate complementation, the limiting-amino-acid problem is already solved at the diet level, and which source supplied the protein barely registers.
These two dials sit on different axes from the source decision, and you satisfy both at once. Biasing protein toward plants for longevity (the mortality section above) and ensuring quality for muscle are not in tension: one is about mortality, the other about muscle, and they meet in a single plan — reach the amount, bias the sources plant-ward, and complement for quality where the plants are low-DIAAS. The plant shift costs a few extra grams or a pairing, not a compromise on either goal.
Who these dials matter for is not uniform. The amount, the timing, and whether supplements help at all depend hard on the stratum.
Who needs more, and who can stop optimizing
The daily number is not one number. Who should raise protein, who has already pulled the lever, and who should leave it alone depend on a person’s age, whether they train, whether they are dieting, and the state of their kidneys. Age, training, dieting, and kidney health are the same effect-modification question seen from four sides — each held on positive evidence rather than mechanism alone, and each resolving differently.
The undershooting older adult: raise the total, spread it across meals
Most older adults eat too little protein, not too much, so the common failure in this stratum is undershoot. PROT-AGE sets the healthy-older target at 1.0-1.2 g per kg of body weight per day, 1.2 for those who are active, and 1.2-1.5 during acute or chronic illness (Bauer et al., 2013). The total is not enough on its own: old muscle needs roughly 25-30 g of protein at each meal to clear the raised anabolic threshold, so the move is to spread protein across breakfast, lunch and dinner rather than back-load it at night (Anabolic Resistance) (Bauer et al., 2013).
For an older adult training for hypertrophy, the muscle-building number — about 1.6 g per kg of total body weight, as measured on a scale — is a floor to reach, not a ceiling to fear. Morton’s 1.62 g/kg break-point is a soft, non-significant knee, fit largely from an almost entirely young sample, and the supplement effect it rests on decays with age (Morton et al., 2017). So the young knee does not transport down to justify aiming low in the old; anabolic resistance means older muscle plausibly needs at least as much per stimulus, not less.
Where undershoot forfeits the muscle a person is training for, biasing the intake toward the upper end of that wide interval is a hedge against missing the target — a Layer-3 decision the person makes under the uncertainty, not a claim that more protein raises the average lean-mass effect.
The older adult who already resistance-trains: training is the whole lever
Here the answer flips, and it is the decision-changing refinement. For a healthy, community-dwelling older adult who already lifts, adding protein or other nutrition on top of the training buys nothing. Choi 2021 (high-tier, 22 RCTs) pooled trials of resistance training plus a nutritional intervention against resistance training alone and found «no significant differences between groups in muscle mass, muscle strength, or physical functional performance» (Choi et al., 2021). The one exception was creatine on lean body mass (n = 4, MD 2.61, 95% CI 0.51 to 4.72), but the between-subgroup test was not significant, so read it as a within-arm signal on a handful of trials, not a demonstrated edge over protein (Choi et al., 2021).
This does not contradict raise and distribute — the two answer different decisions, and the sources themselves say so. Whether an undershooting older adult who reaches adequacy does better is one question; whether extra protein or supplements add anything for someone already training and already nutrient-replete is another. Choi reads the null as headroom in the replete and predicts the stratum where nutrition should still bite: those with a deficiency to repair (Choi et al., 2021).
That predicted stratum is diagnosed sarcopenia, and there the signal is weak and confined to grip, not mass. Song 2023 (moderate-tier, 713 diagnosed-sarcopenic older adults) ran the same contrast and found supplementation «might further enhance grip strength rather than muscle mass» (Song et al., 2023). Hold that grip result loosely: it is borderline (WMD 1.87, 95% CI 0.01-3.74, P = 0.049), high in heterogeneity, sits entirely in the compound protein-plus-vitamin-D arm, and depends on the diagnostic criterion used (Song et al., 2023). It is a between-analysis contrast across two strata, not a within-study interaction test, so it does not lift Choi’s higher-tier null to a benefit.
In a caloric deficit, the target rises to protect lean mass
Dieting raises the requirement. Refalo’s meta-regression finds that «increasing protein intake up to 1.9 g/kgBM/day or 2.5 g/kgFFM/day, on average, is associated with less FFM loss» — the intake at which the trend line crosses from fat-free-mass loss to hold (Refalo et al., 2025). The authors call this exploratory, and two other pools confirm the direction at a lower dose: Wycherley in sedentary overweight and obese dieters and Kim in adults over 50 both resolve a small lean-mass-sparing benefit at roughly 1.2 g/kg (Wycherley et al., 2012) (Kim et al., 2016).
Two bounds matter. Resistance training and a modest (not steep) deficit do more for lean mass than the protein dial, so the training comes first and protein is the adjunct (Refalo et al., 2025). And leaner people gain more per gram — they burn more body protein under restriction — so the target scales inversely with body fat; the obese sit off the studied support and are not evidence that they need less (Refalo et al., 2025). Because undershoot is costlier here than in energy balance, biasing the intake up is the same hedge as above — a decision under the wide interval, not a higher mean effect, and whose loss sets it is the dieter’s.
Healthy kidneys are not a reason to hold protein back
The standing worry is that higher protein damages the kidneys. For normal kidneys it does not. Devries pooled 28 RCTs and, on the change-from-baseline analysis that nets out the arms’ starting difference, found no effect of higher protein on glomerular filtration rate — «a nonexistent or trivial effect of HP consumption on GFR in individuals with normal kidney function» (Devries et al., 2018). The small GFR rise higher protein does produce is benign functional reserve, not damage, so kidney fear is not a reason to hold a healthy person below the muscle or older-adult target.
The one hard flip is established chronic kidney disease. At severe impairment (eGFR below 30, not on dialysis), protein is restricted, not liberalized — a genuine route-(c) contraindication that fires for a defined diseased stratum and does not erode the healthy-adult target (Devries et al., 2018) (Bauer et al., 2013).
The obese target is a named gap, not a scaled-down number
For an obese person the gram target is undefined until the kilogram is named. Morton’s number is grams per kg of total body weight, and total and lean mass diverge about twofold in obesity, so «1.62 g/kg» does not specify a dose until the denominator is stated (Morton et al., 2017). The obese sit off the support of every trial pooled here — Morton, Refalo, Wycherley’s dose-response — so the honest object is a gap, not a scaled-down lean or reference target. Any per-stratum lean scaling is discounted mechanism, sign-uncertain, and never a finding: the obese carry more absolute lean mass, which pushes a per-lean target up, not down.
The evidence is mostly on surrogates — the honest ceiling
Almost every number in this appraisal measures a stand-in, not the thing you care about. Muscle mass, strength, fat-free mass, the glomerular filtration rate, a protein’s DIAAS score — all are surrogates for the outcomes that actually matter, physical function and length of life -> Surrogate Outcomes. The one arm that reaches a hard outcome, protein source and mortality, is observational, not a trial. No randomized trial has ever assigned people a protein target and then measured whether their function held or their lives lengthened, so the loop stays open.
The kidney reassurance rests on the same footing: trials track GFR over weeks, not kidney failure over years, and no protein-to-hard-kidney-outcome trial exists (Devries et al., 2018). And the muscle case carries its own gap — low muscle mass independently predicts mortality, a link de Santana found is not fully explained by muscle strength (de Santana et al., 2021), but that raising mass lowers it is unproven.
What to do — and the question none of it answers
If you resistance-train, aim to reach about 1.6 g of protein per kg of total body weight, built from roughly 25-30 g a meal rather than one large serving — a floor worth hitting, not a ceiling to fear . For a longer life the lever is source, not amount: shift protein-bearing foods from animal toward plant. The stratum decides the rest:
- Older and already lifting: the training is the lever, and added protein or supplements buy little on top of it — don’t chase them, the one exception being a weak grip signal in diagnosed sarcopenia .
- Eating too little, older and undershooting, or in an energy deficit: raise the total and spread it across the day.
- Healthy kidneys: kidney worry is no reason to hold protein below target; established chronic kidney disease is the one place to restrict.
The honest caveat rides on all of it: no trial yet shows that hitting any of these targets changes how you function or how long you live.
Evidence box
Question ’For an adult deciding how much protein to eat and from what source: what is the effect of protein amount, type, timing, and source on each patient-important outcome (muscle mass and physical function, all-cause and cause-specific mortality), what is the dose-response shape, and how does the effect vary by stratum (younger, older, resistance-trained, energy-deficit, renal)?’ Evidence included 15 sources — 8 gold, 4 high, 2 moderate, 1 weak Overall certainty Medium (see Rating Certainty of Evidence) Source-selection note 3 source(s) below the gold evidence bar feed this page: Moore (RCT, moderate); Katsanos (RCT, weak); Song (meta-analysis, moderate). Each labelled by tier; none load-bearing for the core claims. Last updated 2026-09-04 · Independently reviewed: No · Full edit history