Nutrition 01: Protein intake

Short answer. The 1.6 g/kg figure comes from a breakpoint that did not reach statistical significance, with a confidence interval running from about 1.0 to 2.2 g/kg (Morton et al., 2018). Our position is that 1.6 g/kg is a reasonable floor, not a target, and that whether more helps is uncertain, not known to be wasted. Protein is more likely to be over-thought than neglected.

---

What it is

How much protein you eat per day, expressed relative to body mass. The number almost everyone gives is 1.6 to 2.2 grams per kilogram, and almost everyone giving it is citing the same paper, usually without having read how that number was produced.

It was produced carefully. It is repeated carelessly, in the sense that the uncertainty around it gets stripped on the way to you.

The proposed mechanism

Muscle protein is in constant turnover. Training raises the rate of synthesis; dietary protein supplies the amino acids that synthesis runs on and independently stimulates the process. Below some intake, synthesis is substrate-limited. Above it, adding more protein adds nothing because something else has become the constraint.

The question is where that transition sits, and whether it is a transition at all rather than a gradual flattening.

What has actually been tested

The paper everyone cites. A systematic review, meta-analysis and meta-regression pooled 49 studies and 1,863 participants. Protein supplementation produced significant additive gains alongside resistance training: one-repetition maximum increased by 2.49 kg (95% confidence interval, CI, 0.64 to 4.33), fat-free mass by 0.30 kg (0.09 to 0.52), and muscle fiber cross-sectional area by 310 µm² (51 to 570) (Morton et al., 2018).

Those results are solid. Then comes the number.

The breakpoint was 1.62 g/kg/day, at p = 0.079, with a 95% confidence interval running from 1.03 to 2.20.

The threshold underneath most protein advice in existence did not reach conventional statistical significance, and its confidence interval spans a more than two-fold range. The commonly quoted "1.6 to 2.2" is not a recommended range. It is the point estimate and the upper bound of a confidence interval, presented as though it were a target and a ceiling.

Two later analyses used different methods and did not find a plateau there. One modeled the dose-response with splines rather than assuming a breakpoint, on the reasoning that a breakpoint analysis bakes in the assumption that the slope goes flat past some point (Tagawa et al., 2021). Another pooled 66 studies and 2,665 participants and reported by intake category, using subgroups of those trials, not all 66 at once. Below 1.2 g/kg produced no significant effect (g = -0.14, p = 0.35), where g is a standardized mean difference, the gap between groups in standard deviations. Intakes of 1.2 to 1.59 g/kg produced a small one (g = 0.17, p < 0.01), and 1.6 g/kg or above a larger one (g = 0.30, p < 0.01) (Nunes et al., 2022).

That last result is often cited as confirming the 1.6 threshold. It does not. It shows the highest category outperforming the middle one, and the highest category has no upper bound.

The population is the usual problem. Most included trials used untrained participants, and the range of intakes studied was narrow. There is very little data at the intakes people argue about.

Our reading

The evidence supports eating more protein than most people do. It does not support the precision with which the number is stated.

Protein supplementation works. Three meta-analyses agree on that and the effect sizes are consistent. What none of them establishes is a threshold, and the one that came closest reported a non-significant breakpoint with a two-fold interval.

The three analyses used three different methods and the method chose the answer. A breakpoint analysis assumes a plateau exists and finds where it is. A spline model does not assume one. A categorical analysis with an open-topped highest bin cannot detect one. Anyone claiming the literature converges on 1.6 is reading one method's output as though it were the field's conclusion.

So our position is that 1.6 g/kg is a reasonable floor, not a target, and that the interval above it is uncertain, not known to be wasted. If you are at 1.2 and could get to 1.6, the evidence supports doing so. If you are at 1.6 and wondering whether 2.0 does anything, the answer is that the study most often cited to say it does not could not rule out that it does.

And we would not spend much attention above that. The effect sizes here are small in absolute terms. A weighted mean difference (the average gap between groups, weighted by study size) of 0.30 kg of fat-free mass across a training block is real, worth having, and not the difference between progress and no progress. Protein is the nutrition variable most likely to be already handled and most likely to be over-thought.

Where the goals diverge

Building in a surplus. The floor applies and the ceiling is unknown. Erring high costs money and stomach space and, on current evidence, nothing else. The variable with more room to move is training. (→ Loading fundamentals sub-hub)

Dieting. Protein requirements rise in a deficit, and preserving lean mass is the objective rather than adding it. Setting the deficit itself is covered separately. (→ Nutrition 03: BMR, TDEE and the error around them) This literature mostly excluded weight-loss studies, so the numbers above transfer poorly.

Older trainees. Anabolic resistance means the same dose produces less response, and per-meal thresholds appear higher. The daily number is a poor instrument here, as it is for several other variables in this population. (→ Loading fundamentals 08: Warm-up)

Already eating 2 g/kg. Nothing here suggests you should change, and nothing here suggests it is doing more than 1.6 would either. The supplement with a clearer answer is covered next. (→ Nutrition 02: Creatine; Nutrition sub-hub)

The case against

We are criticizing a p-value in a paper whose main findings we accept. The supplementation effects in that analysis were significant and well estimated. Only the breakpoint was not, and the authors reported it transparently. The failure is in the citation chain, not the paper.

A non-significant breakpoint is not evidence of no breakpoint. Our reading emphasizes the uncertainty; someone could equally emphasize that 1.62 remains the best available point estimate. Both readings are defensible.

One author held a supplement industry position that was disclosed three years late. A correction records that a co-author served on the advisory board of a sports supplement manufacturer at the time the paper was written and continued to do so (Correction to Morton et al., 2020). The paper is a meta-analysis of protein supplementation. The pooled result draws on 49 independent trials and is not invalidated by it. The relationship still belongs beside the finding, and the original publication did not carry it.

Per-meal distribution, timing and protein quality are outside this piece. Each has its own literature, and some researchers consider distribution more consequential than the daily total.

Higher-intake evidence is thin in both directions. The claim that more than 1.6 helps is as under-evidenced as the claim that it does not, because almost nobody has run the trials.

What is not known

Whether a plateau exists, and if so where. Three analytical approaches have produced three different-shaped answers from overlapping data.

Whether trained lifters differ from the untrained participants who supplied most of these data.

What the requirement is in a deficit, in older trainees, or at very high training volumes. (→ Loading fundamentals 07: Weekly volume)

Whether daily total is even the right unit, given that per-meal thresholds have been proposed and are measured differently.

Safety

Very high protein intakes are not a concern for healthy kidneys but are for compromised ones. If you have kidney disease or a related condition, speak to a doctor before making large changes.