Inflammation 03: Antioxidant supplements

Short answer. High-dose vitamin C and E, and N-acetylcysteine, blunt adaptive signaling in humans, but pooled trials show no measurable effect on aerobic capacity, endurance performance or lean mass. Our reading is that the money buys nothing measurable for training. If you take them for another reason, moving the dose away from your hardest sessions costs nothing and avoids the only condition under which blunting has been consistently observed.

What it is

Supplements taken to reduce exercise-induced oxidative stress. The category is wide: high-dose vitamin C and E, N-acetylcysteine (NAC), resveratrol and other polyphenols, tart cherry, curcumin, mitochondria-targeted compounds.

They are sold on one story and that story is the same across every bottle. Training generates reactive oxygen species. Reactive oxygen species damage tissue. Mop them up and you recover faster.

The story has a problem that has been known for two decades and has not reached the label. (→ Inflammation sub-hub)

The proposed mechanism

Reactive oxygen and nitrogen species produced during exercise are not only damage. They are signals. They feed into the pathways governing mitochondrial biogenesis, antioxidant enzyme expression, cellular defense and insulin sensitivity, and a substantial body of work now treats them as part of how the cell is told to adapt (Merry and Ristow, 2016).

If that is right, a supplement that scavenges them removes part of the message. And because the mechanism is about scavenging rather than about any particular molecule, it predicts a class effect. Anything with sufficient antioxidant capacity, taken in sufficient dose at the wrong time, should blunt adaptation.

That prediction is the useful thing about this literature, because it is specific enough to fail. It has partly failed, and the pattern of the failure is where the practical answer lives.

What has actually been tested

Vitamin C and E is the best-evidenced case, and it splits by level. Fifty-four people took 1000 mg vitamin C and 235 mg vitamin E daily alongside eleven weeks of endurance training; the supplement blunted the training-induced rise in mitochondrial proteins (Paulsen et al., 2014a). Thirty-two recreationally trained people took the same dose through ten weeks of resistance training; phosphorylation of the growth-signaling proteins p38 MAPK, ERK1/2 and p70S6K was blunted, muscle mass was not, and some strength measures increased less (Paulsen et al., 2014b).

Pooled across randomized trials, the outcome effects are absent (Clifford et al., 2020). In standardized mean differences (SMD, the gap between groups in standard deviations), maximal oxygen uptake was -0.14 (95% confidence interval, CI, -0.43 to 0.15), endurance performance -0.01 (-0.38 to 0.36) and lean mass after resistance training -0.07 (-0.36 to 0.23).

N-acetylcysteine is an independent line pointing the same way. Infusion with NAC attenuated early adaptive responses to exercise in human skeletal muscle (Petersen et al., 2012), and NAC infusion has been shown to modulate exercise-induced changes in insulin sensitivity signaling (Trewin et al., 2015). Different compound, different mechanism of action, same direction. This is what makes the class-effect prediction more than a story about two vitamins.

Then the pattern breaks. Resveratrol supplementation enhanced exercise-induced cellular and functional adaptations of skeletal muscle in older men and women in one trial (Alway et al., 2017), while other trials, primarily in older men, have reported attenuated training responses. A recent review reports two more exceptions (Mănescu et al., 2026). In one trial, a mitochondria-targeted compound augmented the acute response of PGC-1α, a regulator of mitochondrial growth, where blunting was predicted. And chronic polyphenol intake during endurance training shows neutral to supportive signals, with no consistent evidence of blunted mitochondrial signaling, particularly when the dose is positioned away from key sessions.

Where the pattern holds, three things are doing the work. Reviewing this literature, the attenuation is most consistently associated with broad-spectrum scavengers, at chronic high doses, taken close to adaptation-critical sessions. Narrow-spectrum compounds, food-level doses, and timing away from key training do not show the same signal. One review of endurance-focused evidence lands there directly, recommending against chronic combined high-dose vitamin C and E during heavy training while noting NAC may have a place in the days before a competition, not during a training block (Mason et al., 2020).

And it may not be linear. The framing that best fits is hormetic: a moderate redox signal drives adaptation, and both suppressing it and overwhelming it are worse than leaving it alone. If that is the shape, the question is never whether antioxidants are good or bad. It is where on the curve a given dose puts you.

Our reading

The class effect is real but narrower than the class. Two independent compound families, vitamin C plus E and N-acetylcysteine, blunt adaptive signaling in humans. That is not a quirk of one vitamin. But resveratrol, polyphenols and mitochondria-targeted compounds have not behaved the same way, and one of them went the other direction. Whatever is happening is not "antioxidants blunt training."

The variables that predict it are dose, breadth and timing, not the compound. Broad-spectrum, high-dose, taken near the session. That is the profile with the evidence behind it, and it happens to describe exactly how the products are marketed and used.

The signal moves reliably. The outcome mostly does not. This gap is the most abused feature of the literature. Pooled trials show no effect on aerobic capacity, endurance performance or lean mass. Blunted phosphorylation and blunted mitochondrial protein expression are real findings that have not, so far, produced a measurable difference in what people can lift or run.

What we do with this. No demonstrated performance benefit and no reliably demonstrated cost. That is a weaker basis for advice than either camp usually admits.

Our reading is that the money is buying nothing measurable, and that if you take these for a reason unrelated to training, moving the dose away from your hardest sessions costs you nothing and removes the only condition under which blunting has been consistently observed. Food-level intake is not implicated in any of this and nothing here applies to it.

Where the goals diverge

The same pattern of a mechanism outrunning an outcome runs through this whole branch. (→ Inflammation sub-hub)

Building aerobic capacity. The strongest mechanistic case and the worst fit for the practice. The blunted markers were mitochondrial, which is the system you are training.

Building muscle. Pooled lean mass sits on zero and the best-controlled resistance trial found blunted signaling with unblunted growth. Not the intervention that will cost you size, and not doing anything for you either. Anti-inflammatory drugs are the case in this branch where growth itself moved in some trials, in both directions, though most found no difference. (→ Inflammation 04: Anti-inflammatory drugs)

Older trainees. The one outcome-level blunting result is here, and resveratrol has produced findings in both directions in this population. This is the group we would be most cautious about and least confident advising.

Competing soon. Recovery-oriented use around a competition is a different objective from adaptation, and the timing logic reverses. (→ Inflammation 02: Delayed-onset muscle soreness; Inflammation 01: Cold water immersion)

The case against

The strongest counter-evidence is outcome-level and we lead with the null. In thirty-four men aged 60 to 81, total lean mass increased 3.9% on placebo against 1.4% on 500 mg vitamin C and 117.5 mg vitamin E, taken before and after each session over twelve weeks. Rectus femoris thickness increased 16.2% against 10.9%. One-repetition maximum improved 15 to 21% with no group difference (Bjørnsen et al., 2016).

The researchers weight it the other way. Having found no hypertrophy difference, Paulsen and colleagues recommended that healthy young people who train avoid high-dose vitamin C and E. They weighted their mechanistic findings above their outcome findings; we have done the reverse, and they know their data better than we do.

Wide intervals are not evidence of absence. The lean mass interval runs from -0.36 to 0.23. A meaningful negative effect fits inside it.

Our dose-breadth-timing framing is a synthesis, not a tested hypothesis. No trial has manipulated antioxidant breadth or timing as an independent variable and measured adaptation. We are reading a pattern across trials that were not designed to reveal one.

What is not known

Whether timing away from sessions preserves adaptation, which is the most practically useful claim in this piece and rests on inference.

Where the hormetic curve sits, for anyone. The framing is coherent and unquantified.

Whether compounds that have not been tested behave like vitamin C and E or like polyphenols. Most commercial antioxidant products fall in this gap.

Safety

The doses in these trials are well above what food provides; if you take other medications or have a condition affecting absorption or clotting, speak to a doctor before starting.