Plant vs Animal Protein — Does the Source Change the Result?

Protein quality scoring is not marketing. The FAO's Digestible Indispensable Amino Acid Score (DIAAS) grades a protein on whether it supplies every indispensable amino acid in the ratio a human needs, after accounting for how much of it is actually absorbed. On that scale, pork, casein, egg — and, interestingly, potato protein — all score above 100 and count as excellent. Whey and soy clear 75 and count as high quality. Gelatin, rapeseed, lupin, canola, corn, hemp, fava bean, oat, pea and rice all land below 75, in the category that earns no quality claim at all.^[2]

The reason sits in the amino acid profile. When a Maastricht group ran thirty-five commercially available protein powders through the same mass spectrometer, plant isolates averaged 26% essential amino acid by weight against 37% for animal isolates — with human skeletal muscle protein itself sitting at 38%.^[1] Oat, lupin and wheat came in at 21 to 22%, below the WHO/FAO/UNU requirement. Whey topped the list at 43%.

Leucine is the specific amino acid that matters most, because it is the one the muscle's growth machinery actually senses. Plant isolates averaged 7.1% leucine against 8.8% for animal isolates. Lysine and methionine were the bigger shortfalls: 3.6% and 1.0% in plant proteins versus 7.0% and 2.5% in animal proteins.^[1]

But the averages hide an enormous spread, and the spread is the useful part. Corn protein is 13.5% leucine — higher than whey's 11.0%. Potato protein is 8.3% leucine, higher than casein's 8.0% and egg's 7.0%, and it was the only plant isolate in the whole set that met the WHO/FAO/UNU requirement for every essential amino acid. Hemp, at 5.1% leucine, does not meet the requirement for leucine at all.^[1]

The quality numbers predict something measurable. Feed someone a protein and biopsy their muscle over the next few hours, and you can watch the rate at which they build new muscle protein.

In sixty healthy older men, 35 g of casein raised myofibrillar protein synthesis to 0.050 %/h while an identical 35 g of wheat protein hydrolysate managed 0.032 %/h — a real and statistically significant shortfall.^[4] In thirty elderly men, 20 g of soy isolate produced rates lower than 20 g of whey, and not actually distinguishable from eating nothing at all; doubling the soy to 40 g raised the response above zero but still left it below 40 g of whey.^[5] Those two studies are the empirical foundation of the whole "plant protein is inferior" position, and they are not wrong.

They are, however, two protein sources out of hundreds. When the same Dutch group tested the plant proteins that actually score well, the result inverted. Thirty grams of potato protein matched 30 g of milk protein at rest and after exercise.^[6] Thirty grams of corn protein matched 30 g of milk protein exactly — 0.053 %/h versus 0.053 %/h.^[8] A 30 g blend of wheat, corn and pea protein numerically out-performed 30 g of milk protein (0.064 versus 0.053 %/h), although the difference did not reach significance.^[7] Spirulina and chlorella matched mycoprotein.^[16]

If the problem is concentration, the obvious fix is volume. It sometimes works beautifully. In the same wheat protein study, raising the dose from 35 g to 60 g — enough to match the leucine in 35 g of whey — lifted myofibrillar protein synthesis to 0.049 %/h, essentially erasing the deficit.^[4]

It does not always work. Doubling soy protein from 20 g to 40 g in older men narrowed the gap with whey but did not close it, and the authors traced the failure to two things: soy produced a smaller rise in blood leucine, and more of its amino acids were simply burned for fuel rather than built into tissue.^[5] Older muscle is less responsive to an amino acid signal to begin with, so a weaker signal costs more.

The amino acid analysis suggests why one worked and the other didn't, and gives a usable rule. To deliver the 2.7 g of leucine contained in 25 g of whey, you would need 20 g of corn protein, 33 g of potato, 38 g of pea, 40 g of soy, 45 g of wheat or 54 g of hemp.^[1] Blending is more efficient than brute force. Pairing a lysine-poor protein (corn, rice, hemp) with a methionine-poor one (soy, pea) at a 50/50 ratio produces a blend needing only 10 to 90% more protein to cover every essential amino acid, against the two- to four-fold increase a single source would demand.^[1]

Acute muscle protein synthesis is a surrogate. Lean tissue is the outcome. There are now two well-controlled trials that went all the way, and both matched total protein between arms — which is the only way to ask the source question cleanly.

In São Paulo, nineteen habitual vegans and nineteen habitual omnivores were brought to 1.6 g/kg/day of protein (soy isolate for the vegans, whey for the omnivores) and trained twice weekly for twelve weeks. Leg lean mass rose 1.2 ± 1.0 kg in the vegans and 1.2 ± 0.8 kg in the omnivores. Vastus lateralis cross-sectional area rose 2.2 ± 1.1 versus 2.8 ± 1.0 cm². Type II fibre area rose 921 ± 458 versus 844 ± 638 µm². Leg press one-rep max rose 97 ± 38 versus 117 ± 35 kg. There was no significant difference on any measure.^[13]

In Exeter, twenty-two young adults trained five days a week for ten weeks on roughly 2 g/kg/day of protein, either omnivorous or exclusively non-animal with mycoprotein as the backbone. Lean mass rose 2.6 ± 1.1 kg on the omnivorous diet and 3.1 ± 2.5 kg on the vegan one. Thigh muscle volume rose 8.3% in both groups — to the decimal. Fibre cross-sectional area rose 33% and 32%. Strength rose comparably across multiple lifts.^[11]

Honesty requires printing the trial that disagrees. When the Maastricht group stopped testing isolated powders and tested whole meals, in sixteen adults aged 65 to 85, a meal built around 100 g of lean beef produced about 47% higher postprandial muscle protein synthesis than an isonitrogenous, isocaloric whole-food vegan meal — 0.052 versus 0.035 %/h.^[9] That is the single strongest result in favour of animal protein in the modern literature, and it was obtained in exactly the population with the most to lose.

It is not the last word. A crossover trial in thirty-four community-dwelling adults averaging 72 years put each person on a ten-day controlled vegan diet and a ten-day isocaloric, isonitrogenous omnivorous diet, and measured integrated muscle protein synthesis across the whole period rather than across one meal. The rates were 1.23 %/day on the vegan diet and 1.29 %/day on the omnivorous diet — not significantly different.^[10] The difference between the two studies is instructive: one meal, built around a single plant source, loses; ten days of a varied plant-based diet does not. A single meal is not a diet.

A 2026 systematic review pooled eighteen randomised trials in 1,893 adults that compared plant with animal protein supplementation for at least six months. It found no significant difference in lean body mass, fat mass, total body mass, upper- or lower-body strength, gait speed, chair stand, timed-up-and-go, lipids, blood pressure, fasting glucose, insulin or HOMA-IR.^[14]

That reads like a settled answer. It is worth looking at the confidence intervals before treating it as one. The lean body mass estimate pooled only seven trials and 365 participants, and came out at a standardised mean difference of 0.26 with a 95% confidence interval running from −0.58 to 0.90 — an interval wide enough to contain a moderate advantage for either side. Heterogeneity was 92.3%. Lower-body strength was worse still: three effect sizes, an interval from −1.07 to 6.14, heterogeneity 98.6%. Only gait speed was clean. The authors also note evidence of small-study effects on most outcomes, and six of the eight are employed by a nutrition company.^[14] The correct reading of this meta-analysis is not "the sources are equivalent" but "the long-term trials are too few, too small and too different from each other to detect a difference if one exists."

Every trial that found no difference achieved that result under a specific condition: total protein was adequate and matched. 1.6 g/kg/day in São Paulo, 2 g/kg/day in Exeter, isonitrogenous diets in Wageningen. None of them tested whether a typical plant-based diet, eaten the way people actually eat it, reaches those numbers — and that, rather than amino acid chemistry, is the real risk. If you want the arithmetic behind those targets, our piece on how much protein per day works through the evidence, and the companion on protein timing and frequency covers how to spread it. Tracking what you actually eat for a fortnight is the cheapest way to find out whether you are near 1.6 g/kg or nowhere near it; our free Peak Nutrition tracker sets targets from your own labs.

The second adjustment is per-meal. The dose that reliably triggers muscle protein synthesis is roughly what 25 to 30 g of a high-quality animal protein delivers. Hitting the same leucine from a single plant source means eating meaningfully more of it, so a plant-based meal aimed at muscle should be built around 30 to 40 g of protein rather than 20 to 25 — and preferably around the sources that score well. Soy, pea, potato, corn and mycoprotein all have evidence behind them; hemp, oat and lupin do not.

The third is variety. Combining complementary sources across the day covers the lysine and methionine shortfalls at far lower total intake than loading up on any one of them, and the varied-diet result in older adults suggests this is not a theoretical nicety.^[1][10] Lifting remains the signal, in any case: across 49 trials, training itself added 27 kg to a one-rep max while the protein supplement added 2.5.^[15] For how that fits a wider programme, see our performance guide.