Who Extra Protein Actually Helps

Somewhere in most kitchens there is a tub with a scoop in it. The label promises muscle, and the internet promises more muscle per gram. What almost nobody does is check the number on the tub against the number on the bathroom scale — and that piece of arithmetic, not the powder, turns out to decide whether the tub does anything at all.

This spring a Dutch team published the cleanest test of the question I have read in a while, and the headline result was a shrug. Then they split the participants by how much protein they were already eating, and the shrug became the most useful protein finding of the year.

A trial that added protein to lifting and found nothing

The trial recruited 295 frail community-dwelling older adults — mean age 73.9, sixty-nine percent women — and put every single one of them on the same programme: twice-weekly full-body resistance training, taken to muscle failure, for twelve weeks. Half of them also received dietary counselling and 20 g of whey protein a day, aimed at a total intake of 1.2 to 1.5 g per kilogram of bodyweight.

The protein arm actually hit its mark, which is rarer than it sounds. By follow-up they were eating 0.4 g/kg/day more than the training-only group (95% CI 0.3–0.5, p<0.001). A real, verified separation — not a good intention that evaporated by week three.

And the primary outcome barely moved. Leg-press one-rep max improved by 4.2 kg more in the protein group, with a confidence interval of −0.3 to 8.8 kg and a p-value of 0.069 — a range that still contains zero. Lean mass in the arms and legs: no difference. Physical performance: no difference. Two hundred and forty-four people analysed, a verified extra 20 g a day, twelve weeks of supervised lifting, and the protein added nothing the trial could confidently point at.

Then they looked at who had arrived short

Baseline intake in both groups averaged about 1.0 to 1.1 g/kg. Which is to say: a good number of these people were already eating a respectable amount of protein before anyone handed them a shaker. So the researchers asked the obvious follow-up. Did the ones who came in short do better?

They did, and not subtly. Among participants eating under 1.2 g/kg at baseline, the protein group gained 6.4 kg more leg-press strength (95% CI 1.3–11.4, p=0.013). Among those at or below 0.8 g/kg — roughly the long-standing 0.8 g/kg reference intake, the official floor — the gap widened to 10.9 kg (95% CI 1.5–20.4, p=0.025). The formal test of whether baseline intake changed the effect came out at p=0.038.

Same supplement. Same training. Same twelve weeks. The benefit was not spread thinly across everybody. It was concentrated almost entirely in the people who had been under-eating.

Hold that at arm's length for a moment, because the authors do. Baseline intake came from a three-day food diary that participants filled in themselves, and food diaries flatter their authors. The subgroups were small, the comparisons were many, and the analysis was explicitly exploratory — built to generate the hypothesis, not to close it. The cohort was also fairly functional for a group labelled frail, which leaves less headroom to improve. None of that makes the pattern disappear. It makes it a strong lead rather than a verdict.

Protein behaves like a floor, not a dial

What the trial found in miniature, the meta-analytic literature has been describing at scale for years, and almost nobody reads it that way.

Pool 49 randomised trials and 1,863 people, as a 2018 British Journal of Sports Medicine review did, and protein supplementation on top of resistance training produces a genuine but modest average effect: 2.49 kg on one-rep max (95% CI 0.64–4.33) and 0.30 kg of fat-free mass (95% CI 0.09–0.52). Two and a half kilos and three hundred grams. Real, and rather smaller than a tub's worth of graphic design implies.

More to the point, that average conceals a shape. A 2022 dose–response analysis in Sports Medicine – Open, covering 59 training studies and 2,440 participants, found strength climbing by 0.72% for every additional 0.1 g/kg/day (95% CI 0.40–1.04) — up to about 1.5 g/kg/day, and then nothing further. A slope, then a flat line.

Which is what a floor looks like in data. Protein does not scale with intake so much as it stops being the constraint at some point. Below the bend, the nutrient is what limits you; above it, something else is. Adding more of a resolved constraint is not a marginal gain — it is zero, dressed as diligence.

The famous threshold is a smear, not a line

You will have seen 1.6 g/kg quoted as though it were etched somewhere. It comes from that 2018 review, which put the breakpoint for further fat-free-mass gains at 1.62 g/kg/day. What almost never travels with the number is its confidence interval: 1.03 to 2.20 g/kg/day.

That is not a line in the sand. It is a smudge more than a full gram per kilogram wide — for an eighty-kilo adult, the difference between 82 g and 176 g a day. The 2022 analysis landed on 1.5; the 2026 trial used 1.2 as its target. The honest reading of three converging estimates is not a precise number but a direction: somewhere in the region of 1.2 to 1.6 g/kg/day, the curve flattens, and the exact spot is person-shaped.

This matters because precision theatre is how a floor gets sold as a dial. If the threshold were 1.62, then 1.7 would be a strategy. Given the real interval, "am I roughly in the zone" is the only question the evidence can answer — and for most people it answers it with a yes or a fairly obvious no.

The supplement ceiling is lower than the tub implies

There is a second ceiling, in the powder itself. A Bayesian network meta-analysis published in Frontiers in Nutrition on 6 August 2026 — 36 trials, 1,552 participants, 89 trial arms — modelled the dose–response for supplemental protein specifically, and found the benefit for lean body mass plateauing at roughly 30–35 g a day of whey. In adults 65 and over the plateau shifted a little rightward, to about 35–40 g, consistent with older muscle needing a larger stimulus to respond. Casein and milk protein looked similar, at around 25–35 g, with wider uncertainty.

Two scoops, in other words, is generally the top of the useful range rather than the bottom. And that analysis carries its own caveats worth stating: 72% of the pooled participants were men, and only nine of the 36 trials studied adults over 65 — the very group in whom the question matters most.

For context on the other end of the argument, we have written about the case for eating less protein and where it stops applying, and about how much protein active and older adults actually need. The two posts do not contradict each other; they describe opposite ends of the same curve.

The awkward part: the lifting was the active ingredient

Every participant in the 2026 trial trained. Both arms. Twice a week, full body, to failure, supervised, for twelve weeks. The protein was the variable; the training was the constant. And the constant is where the strength came from.

There is a small gift buried in the methods, too. Participants were assigned training intensities spanning 20% to 80% of their one-rep max, grouped into four bands from light to heavy — and the researchers found no meaningful interaction between load and the protein effect. Read that as what it is: across a fourfold range of load, the programme still worked, provided sets were taken close to failure. If you have been avoiding starting because you do not know your correct percentage, that turns out not to have been the obstacle.

Protein contributes to the growth and maintenance of muscle mass — that much is an authorised, uncontroversial nutritional fact. But it contributes to a process that something else has to start. Nutrition is permissive; the training is causal. Our guide to strength after 40 covers the part that actually does the work, and the difference between being active and being trained covers why steps do not substitute for it.

What to do with this in the next week

One number decides which half of this article applies to you, and you can have it by Friday.

Write down what you eat for three ordinary days — not three virtuous ones — and total the protein. Divide by your bodyweight in kilograms. That is your g/kg. Most people are surprised in one direction or the other, and the surprise is the whole point: nobody's intuition about their own protein intake is reliable, which is precisely why supplement marketing works on everyone equally.

And if you are not lifting, the number is moot. Protein without a mechanical stimulus is a nutrient with nothing to build. Start there; count later.

The bottom line

Add 20 g of protein a day to a supervised strength programme in older adults and, on average, nothing measurable happens. Do the same for the people who were eating under 1.2 g/kg — and especially under 0.8 — and something clearly does. Three separate lines of evidence, from a 2026 trial and two meta-analyses, describe the same shape: a slope that flattens somewhere between roughly 1.2 and 1.6 g/kg/day, with the honest uncertainty on that bend running wider than a gram.

So protein is not a lever you pull harder. It is a threshold you clear once and then stop thinking about. The reason this feels unsatisfying is the reason it is worth knowing: it converts an ongoing purchase into a one-off arithmetic problem, and hands the rest of the budget back to the training, which is where it was always doing the work. Use numbers to correct fantasy, not to replace effort.