Antioxidants: What They Actually Do, and Who Actually Benefits

Antioxidants are molecules that slow or prevent the oxidation of other molecules. The useful version of that answer is less tidy. Your body needs a certain amount of oxidation to function at all, most dietary antioxidants do not work by neutralizing radicals, and high-dose antioxidant supplements have failed repeatedly — occasionally harmfully — in very large trials. The goal is balance, not annihilation.

What Are Antioxidants, in Plain Terms?

An antioxidant is anything that gets oxidized so something more important doesn't. It donates an electron and absorbs the hit. That is the whole chemistry.

The word covers four categories that most articles blur together:

• Direct scavengers. Vitamin C in water-based compartments, vitamin E in membranes. One molecule, one radical.
• Enzymes. Superoxide dismutase, catalase, glutathione peroxidase. Catalytic, so one molecule handles millions of reactions. This is where real capacity lives.
• Cofactors. Selenium, zinc, copper, manganese, iron. They do nothing alone. They sit in the active sites of the enzymes above.
• Signaling phytochemicals. Polyphenols and isothiocyanates, filed under "antioxidants" for historical reasons. As you'll see, that is largely a misnomer.

What Are Free Radicals and Reactive Oxygen Species?

"Free radical" and "reactive oxygen species" are not synonyms. A radical has an unpaired electron. Several of the most important ROS have none — hydrogen peroxide and peroxynitrite are both non-radicals, and hydrogen peroxide is the most important signaling oxidant in the body.

Species	Radical?	Behavior	Biological role
Superoxide (O₂•⁻)	Yes	Charged, short-lived, crosses membranes poorly	Precursor to most other ROS; destroys nitric oxide
Hydrogen peroxide (H₂O₂)	No	Uncharged, diffusible, stable; crosses membranes via aquaporins	The principal redox signaling molecule
Hydroxyl radical (•OH)	Yes	Reacts with almost anything within a nanometre	Pure damage. No signaling role, no enzyme defends against it
Peroxynitrite (ONOO⁻)	No	Formed when nitric oxide meets superoxide	Nitrates tyrosines, oxidizes thiols, destroys BH₄
Singlet oxygen (¹O₂)	No	Excited oxygen from UV and photosensitization	Skin and eye damage; quenched by carotenoids
Lipid peroxyl radicals (LOO•)	Yes	Propagate chain reactions through membranes	The chain vitamin E exists to terminate

Where they come from. Mitochondrial electron transport leaks electrons onto oxygen, mainly at complex I and complex III, producing superoxide as a by-product of making ATP — the subject of our guide to how your body actually makes energy. The NADPH oxidase (NOX) family is different: seven isoforms whose dedicated job is producing ROS on purpose, most famously NOX2 in neutrophils. Xanthine oxidase generates superoxide during purine breakdown. And nitric oxide synthase, when tetrahydrobiopterin runs short, "uncouples" and makes superoxide instead of nitric oxide.

One footnote: the old claim that 1–2% of consumed oxygen becomes superoxide came from isolated mitochondria under artificial conditions. Modern estimates in intact tissue are far lower.

Why ROS Are Not Simply "Bad": Redox Signaling and Hormesis

This is the most important idea on this page, and almost no consumer article says it.

In 2020, Helmut Sies and Dean Jones described ROS in Nature Reviews Molecular Cell Biology as pleiotropic physiological signalling agents, formalizing two terms now standard in the field. Oxidative eustress is low-level ROS engagement in normal signaling, supporting adaptation to a changing environment. Oxidative distress is elevated formation causing molecular damage. Same chemistry, different concentration, opposite consequence.

The mechanism is precise. Hydrogen peroxide oxidizes specific low-pKa cysteine residues to sulfenic acid, forming reversible on/off switches. Peroxiredoxins act as sensors and relays. NRF2 and NF-κB sit downstream and run the cell's stress-response programs.

Which is why several processes you very much want require ROS:

• Immune killing. The NOX2 respiratory burst is how neutrophils kill bacteria and fungi. People born unable to assemble NOX2 have chronic granulomatous disease and suffer recurrent, life-threatening infection. That is the strongest evidence available that superoxide is not merely a pollutant.
• Insulin signaling. The canonical account is that insulin binding triggers a local burst of hydrogen peroxide, which reversibly inactivates the phosphatase PTP1B and lets receptor signaling proceed.
• Exercise adaptation. The transient ROS pulse from muscle contraction drives AMPK, SIRT1, PGC-1α and NRF2 signaling — the pathways producing mitochondrial biogenesis and, ironically, upregulation of your own antioxidant enzymes.

This is hormesis: a modest stressor producing an adaptive response larger than itself. The goal is not fewer radicals. It is redox balance and signal fidelity.

Why Did the Big Antioxidant Supplement Trials Fail?

They failed badly, and honesty about this is non-negotiable. Harvard's Nutrition Source reaches the same conclusion we do: antioxidant supplements have shown little disease-prevention benefit. We agree. The details matter more than the headline.

Trial	Population and intervention	Result
ATBC (NEJM, 1994)	29,133 male smokers, mean age 57; 50 mg alpha-tocopherol and/or 20 mg beta-carotene daily, mean 6.1 years	Beta-carotene increased lung cancer — 474 cases versus 402 in controls. Supplementation "does not prevent lung cancer in older men who smoke."
CARET (NEJM, 1996)	18,314 heavy smokers and asbestos-exposed workers; 30 mg beta-carotene plus 25,000 IU retinyl palmitate daily	Stopped 21 months early. Lung cancer up 28%, all-cause mortality up 17%. Six-year follow-up found excess risk persisted in women (all-cause RR 1.37).
SELECT (JAMA, 2011)	34,887 healthy men; 400 IU/day all-rac-alpha-tocopheryl acetate and/or 200 µg/day L-selenomethionine	Vitamin E alone increased prostate cancer: HR 1.17 (99% CI 1.004–1.36), an excess of 1.6 cases per 1,000 person-years. Selenium alone: no effect.
Bjelakovic Cochrane review (2012)	78 randomized trials, 296,707 participants	Among 56 low-bias trials (244,056 participants), antioxidants increased mortality: RR 1.04 (1.01–1.07). Beta-carotene RR 1.05; vitamin E RR 1.03. Vitamin C and selenium not significant.

Now the part usually left implicit. Every one of those trials tested isolated compounds, at pharmacological doses, given chronically, to people who were not deficient. It failed for the reason the redox biology predicts: flooding a signaling system with one reducing agent degrades the signal.

What they do not indict is food. A meta-analysis of seven prospective cohorts covering 178,657 people found higher dietary polyphenol intake associated with 7% lower all-cause mortality (HR 0.93, 0.91–0.95). Observational, and confounded by everything that correlates with eating plants — but the direction is the opposite of the megadose trials.

Can Antioxidant Supplements Blunt Exercise Adaptation?

Sometimes, yes — the most useful thing an active person can take from this literature.

In 2009, Ristow and colleagues reported in PNAS that vitamin C (1,000 mg/day) plus vitamin E (400 IU/day) during a four-week training program prevented the exercise-induced improvement in insulin sensitivity, and prevented induction of PGC-1α, PPARγ and the endogenous enzymes SOD1, SOD2 and GPx1 in human muscle. The supplements did not merely fail to help. They removed a benefit training otherwise delivered.

Paulsen and colleagues, in The Journal of Physiology (2014), gave 1,000 mg vitamin C and roughly 235 mg vitamin E across 11 weeks of endurance training; COX-IV protein and cytosolic PGC-1α were attenuated, though VO₂max improvement was not. The honest counterweight: the same group's strength study found altered p70S6K and MAPK signaling but no reduction in lean mass or cross-sectional area gains. Not every study reproduces the effect.

The rule that survives: do not take gram-doses of isolated vitamin C and E around the sessions you are training to adapt from. Short, food-first polyphenol exposure during congested competition, travel or back-to-back loads is a different proposition — readiness rather than adaptation, a theme in our guide to minerals in recovery.

Your Own Antioxidant Enzymes — and Their Mineral Cofactors

Most articles skip the part where your body already runs an enzymatic defense system that dwarfs anything you can swallow. It is catalytic rather than stoichiometric, which is why it wins on capacity by orders of magnitude.

• Superoxide dismutase. SOD1 in cytosol (copper, zinc), SOD2 in the mitochondrial matrix (manganese), SOD3 in the vessel wall. All convert superoxide to hydrogen peroxide near the diffusion limit.
• Catalase. A heme-iron enzyme, largely peroxisomal, among the fastest enzymes known.
• Glutathione peroxidases (GPX1–4). Selenocysteine enzymes. GPX4 uniquely reduces lipid hydroperoxides inside membranes; its failure defines ferroptosis.
• Peroxiredoxins handle most low-level hydrogen peroxide and double as signal relays. Thioredoxin and thioredoxin reductase (also selenoproteins) plus glutaredoxin reset oxidized protein thiols.
• Glutathione, at 1–10 mM inside cells, is the dominant redox buffer. The GSH/GSSG ratio is the standard readout of redox state.

Which brings a distinction worth stating flatly: selenium, zinc, copper and manganese are not antioxidants. They neutralize nothing. They are the metal centers that let your own enzymes function.

Mineral	Enzyme it enables	Notes
Selenium	Glutathione peroxidases, thioredoxin reductases	The human genome encodes 25 selenoproteins. RDA 55 µg/day, UL 400 µg/day.
Zinc and copper	SOD1 (cytosolic), SOD3 (extracellular)	Zinc also restrains NADPH oxidase. High-dose zinc can induce copper deficiency.
Manganese	SOD2 (mitochondrial)	The primary defense inside the organelle producing most superoxide.
Iron	Catalase (heme)	Double-edged. Free, unbound iron drives Fenton chemistry, producing hydroxyl radicals.

This is the bridge back to trace minerals, and a real one. A selenium or manganese shortfall does not announce itself as "low antioxidant status." It shows up as a slightly less capable enzyme system, quietly. See also nutrient cofactors and synergy, what trace minerals actually are, and the soil depletion data. Magnesium belongs here too — it is required for glutathione synthesis, one of many jobs in our magnesium guide.

NRF2 and KEAP1: the master switch

Normally a protein called KEAP1 tags the transcription factor NRF2 for destruction as fast as it is made. Human KEAP1 carries 27 cysteine residues acting as chemical sensors — Cys151 triggers many inducers, while Cys273 and Cys288 are required for basal repression. When an oxidant or electrophile modifies them, the disposal machinery stalls, newly made NRF2 reaches the nucleus, pairs with small Maf proteins, and switches on the antioxidant response element: NQO1, heme oxygenase-1, the rate-limiting enzymes of glutathione synthesis (GCLC, GCLM), glutathione S-transferases, thioredoxin, ferritin. NRF2 activation makes you build more of your own defenses.

One balancing note: constitutive NRF2 activation is oncogenic — KEAP1 and NFE2L2 mutations drive chemoresistance in lung and other cancers. NRF2 is a thermostat, not a dial to turn up.

This is the least glamorous and most defensible reason a broad mineral formula includes selenium, zinc, copper and manganese at all. Current supplies each of them at food-range amounts — not as antioxidants, but as the cofactors your own antioxidant enzymes are built from.

How Do Polyphenols Actually Work? (It Isn't Scavenging)

The standard story — polyphenols are powerful antioxidants that mop up free radicals in your body — is arithmetically impossible, and the field has known it for years.

Dietary polyphenols reach plasma concentrations of roughly 0.1–2 µM. The Linus Pauling Institute reports median peak concentrations near 0.9 µM from food and 3.2 µM from supplements. Glutathione sits at 1–10 mM inside cells and urate at 3–7 mg/dL in plasma — three to four orders of magnitude higher. A compound at a thousandth the concentration of the incumbent cannot be doing the incumbent's job by competition. The Institute puts it plainly: flavonoids are effective radical scavengers in a test tube, but their direct antioxidant contribution in the body is likely negligible, and their real effects come from modulating cell-signaling pathways.

So what are they doing? The current model, set out in a 2020 review in Antioxidants, is counterintuitive:

1. Polyphenols autoxidize at the cell membrane, generating superoxide and then hydrogen peroxide at nanomolar to low-micromolar levels — functionally, pro-drugs for hydrogen peroxide.
2. At roughly 10–100 nM, that peroxide inhibits protein tyrosine phosphatases by oxidizing their catalytic cysteines, increasing phosphorylation and producing adaptive signaling. Eustress.
3. At roughly 0.1–5 µM, it activates NRF2, and the cell upregulates its own SOD, catalase and glutathione systems. This is the real antioxidant effect — indirect, enzymatic, and lasting far longer than the polyphenol's plasma half-life.
4. Above about 5 µM: distress and cytotoxicity. Which is exactly why megadosed polyphenol extracts are not automatically better.

Read that sequence again and the megadose failures stop looking mysterious. A mild pro-oxidant signal induces a durable enzymatic defense. A large direct-reducing dose flattens the signal instead.

Bioavailability, and why your gut bacteria matter

Most polyphenols are absorbed poorly. Resveratrol is the textbook case: roughly 75% absorbed, about 1% bioavailable as the parent compound. Much of what you eat reaches the colon intact, where microbiota cleave it into smaller, better-absorbed, often more bioactive metabolites. Ellagitannins from pomegranate, berries and walnuts become ellagic acid and then urolithins; urolithin A induces mitophagy, the clearance of damaged mitochondria. A randomized trial reported improved muscle endurance and mitochondrial biomarkers in middle-aged adults, and a 2025 trial in trained male distance runners found enhanced recovery but no performance improvement. Not everyone produces urolithins — it depends on the microbiome, as does equol from isoflavones. In several cases the metabolite, not the parent compound, is what circulates.

Why Did the USDA Withdraw the ORAC Database?

In 2012 the USDA Agricultural Research Service removed its ORAC database from the web. The reasons were specific: ORAC values have "no relevance to the effects of specific bioactive compounds," and were "routinely misused by food and dietary supplement manufacturing companies to promote their products." The agency added that non-antioxidant mechanisms, still undefined, may be responsible for the benefits of polyphenol-rich foods.

Why it failed is clear once you understand the previous section. ORAC measures how well a compound quenches a peroxyl radical in a cuvette. So do TEAC, FRAP and DPPH — all measure reducing power against synthetic radicals in buffer or solvent. None account for absorption, metabolism, or the fact that the dominant real mechanism is enzyme induction. A compound can score enormously in the tube and never reach a blood concentration where that score means anything. Here is what researchers measure instead:

Marker	What it measures	Status
F₂-isoprostanes (8-iso-PGF2α)	Free-radical-catalyzed peroxidation of arachidonic acid	The most specific and stable in-vivo lipid peroxidation marker; widely called the gold standard
8-OHdG / 8-oxo-dG	Oxidative DNA damage	Useful, but results vary by assay
GSH/GSSG ratio	Global cellular redox state	Meaningful but very sensitive to sample handling
Protein carbonyls, 4-HNE	Protein oxidation and aldehyde adducts	Validated in human altitude and muscle-wasting work
MDA / TBARS	Lipid peroxidation	Non-specific. TBARS reacts with many aldehydes and sugars

A caution applying to all of them: day-to-day and diurnal variability in these biomarkers runs at coefficients of variation of 7.4% to 31.2%. A single before-and-after measurement is close to meaningless.

The Antioxidant Network: Why a Spectrum Beats a Megadose

1. Compartments. Vitamin C works in plasma, cytosol and extracellular fluid; vitamin E and carotenoids in membranes and lipoproteins; glutathione in cytosol and mitochondria; SOD3 on the endothelial surface. No single molecule covers all of them.

2. Species specificity. SOD handles superoxide. Catalase, glutathione peroxidase and peroxiredoxins handle hydrogen peroxide. GPX4 handles lipid hydroperoxides. Carotenoids quench singlet oxygen. Nothing scavenges the hydroxyl radical — it reacts too fast. You can only prevent its formation.

3. Mechanism. Direct scavenging, metal chelation, NRF2-mediated enzyme induction, NF-κB modulation and substrate provision are five different jobs. Vitamin C does the first, sulforaphane the third, selenium the fifth.

4. Regeneration chemistry. A used antioxidant is itself a radical. Vitamin E, having terminated a lipid chain reaction, becomes the alpha-tocopheroxyl radical — and ascorbate regenerates it at the lipid–water interface. Glutathione and NADPH then regenerate ascorbate; the pentose phosphate pathway regenerates NADPH. Break one link and everything downstream stops resetting. That is chemistry, not marketing.

Superoxide Destroys Nitric Oxide

If you take one mechanism away from this page, make it this one.

Nitric oxide and superoxide react to form peroxynitrite at a rate approaching the diffusion limit — Kissner and Koppenol measured it at roughly 1.9 × 10¹⁰ M⁻¹s⁻¹. Superoxide dismutase works at about 2 × 10⁹ M⁻¹s⁻¹. So nitric oxide reacts with superoxide roughly an order of magnitude faster than your own enzyme can intercept it. Wherever both exist, superoxide preferentially consumes nitric oxide. Rafael Radi's 2022 review in Physiological Reviews calls this the superoxide radical switch in the biology of nitric oxide.

Then it compounds, because peroxynitrite oxidizes tetrahydrobiopterin, the cofactor eNOS needs to stay coupled. Lose enough BH₄ and eNOS transfers electrons to oxygen instead of arginine, producing superoxide rather than nitric oxide — destroying more NO and oxidizing more BH₄. A closed loop.

The implication: you cannot supplement your way to more nitric oxide while producing superoxide faster than you can dispose of it. Lowering oxidative burden is not a separate benefit from supporting nitric oxide. It is a prerequisite. We take this apart in our guide to nitric oxide and circulation. One number there shows hormesis in a human vascular endpoint: a dose-response meta-analysis in Food & Function found cocoa flavanols improved flow-mediated dilation by 1.17% (95% CI 0.76–1.57), with an inverted-U peak near 710 mg total flavanols. More was worse. In humans. Measured.

Who Actually Benefits? A Population-by-Population Look

This is the question most authority pages don't answer, and the only one most readers have. The honest answer varies enormously by who you are.

Population	Evidence of elevated oxidative burden	Evidence intervention helps
Older adults	Strong (inflammaging, declining NRF2 responsiveness)	Moderate
Low fruit and vegetable intake	Strong intake data, indirect biomarkers	Moderate, largely by inference
Smokers and vapers	Strong for smoking	Vitamin C status yes; beta-carotene harmful
Metabolic dysfunction, obesity	Strong, dose-responsive with BMI	Moderate on biomarkers
Air pollution exposure	Strong mechanistically	Moderate (broccoli sprout trials)
Athletes, high training load	Moderate, transient by design	Moderate for recovery; risk of blunting adaptation
Recovery from illness	Moderate (vitamin C turnover)	Moderate for status restoration
Altitude and high UV	Moderate to strong	Weak — exposure real, intervention unproven
High stress, poor sleep	Weak and inconsistent	Weak
Shift workers	Contested — several null studies	Weak

Older adults

The strongest case, though not for the usual reason. The naive free-radical theory of aging has not survived the data. The mature framing is loss of redox signaling fidelity and mitochondrial quality control, plus inflammaging: elevated pro-inflammatory cytokines, cellular senescence, persistent NF-κB activation, and impaired NRF2-mediated antioxidant defense. NRF2 responsiveness declines with age, so the induction pathway phytochemicals use is precisely the one weakening.

Anyone eating few fruits and vegetables — by far the largest group

According to CDC surveillance data from 2019, only 12.3% of US adults met fruit intake recommendations and 10.0% met vegetable recommendations. Vegetable intake was worst among adults aged 18–30 (7.1%) and in the lowest income group (6.8%), ranging by state from 5.6% in Kentucky to 16.0% in Vermont. Average US vitamin E intake runs 7.2–11.5 mg/day against an RDA of 15 mg, and over half of US adults fall short on dietary magnesium. This unglamorous case is the one we find most persuasive — a theme in the daily foundation.

Smokers, vapers, and people breathing polluted air

Smokers have measurably lower plasma vitamin C, A and E, and elevated MDA. Smoking increases vitamin C turnover enough that the US RDA is formally raised by 35 mg/day for smokers — one of very few dietary reference values adjusted for an oxidative-stress exposure. Fine particulate matter carries transition metals and quinones that generate ROS directly, and PM2.5 exposure is linked to oxidative stress, DNA damage and chronic inflammation.

The caveat this section must carry: smokers and asbestos-exposed workers are exactly the population in whom beta-carotene supplementation increased lung cancer. If you smoke, do not take beta-carotene supplements. Food carotenoids are a different matter. For wildfire smoke and vaping specifically, the mechanism is expected but we found no strong human antioxidant-intervention evidence — plausible, not proven, and our post on wildfire smoke and your body stays on the same side of that line.

Athletes, metabolic dysfunction, and everyone else

Two things are true of athletes at once. Higher physical activity is associated with lower urinary 8-OHdG and 8-iso-PGF2α — training improves your endogenous defenses. And chronic high-dose isolated vitamins around adaptation-critical sessions can interfere with that. The best-evidenced use case is congested competition and travel, when readiness beats adaptation.

In metabolic dysfunction the human data is unusually clean. Serum 8-iso-PGF2α rose stepwise with BMI: 219 pg/mL in non-obese participants, 246 in grade I obesity, 431 in grade II–III obesity. An underused point: glycemic excursions, not just mean glucose, associate with urinary isoprostanes, in normal glucose tolerance as well as diabetes.

Infection increases vitamin C turnover, and prophylactic adequacy takes 100–200 mg/day, as covered in our complete guide to vitamin C. Beyond that, be careful. For high stress and poor sleep the mechanism is plausible but human biomarker evidence is inconsistent — the mineral side, in why stress burns through your minerals, rests on firmer ground. For shift work, a 2023 scoping review in Environment International found associations "albeit with mixed results," and recent studies of rotating-shift nurses found no significant difference in total antioxidant status. Most supplement articles present shift work as settled. It is not. At altitude, the Caudwell Xtreme Everest expedition found protein carbonyls and 4-HNE correlated with tissue loss — yet a review of 19 antioxidant studies for acute mountain sickness found only a non-significant trend. Real exposure does not make the intervention proven.

For how the wider category handles the mineral cofactors specifically — most skip them entirely — see the 2026 electrolyte powder comparison.

The Seven Botanicals, Graded Honestly

Green tea, green coffee, turmeric, blueberry, cherry, broccoli and kale appear constantly in polyphenol formulas. Here is what each actually has behind it.

Green tea and EGCG

Catechins make up 30–42% of the solid weight of brewed green tea. Mechanism: NRF2 activation, NF-κB and MAPK modulation, iron chelation, and — after gut microbial ring fission — circulating gamma-valerolactones rather than intact EGCG. Observational meta-analyses link high consumption to roughly 33% lower cardiovascular mortality, which is impressive and thoroughly confounded. On cancer the Linus Pauling Institute is blunt: current evidence does not support a role for tea consumption in preventing most cancers in humans. EFSA concluded in 2018 that 800 mg or more of EGCG daily from supplements significantly raises serum transaminases; brewed tea at 90–300 mg/day is considered safe. Grade: strong mechanistically, moderate for vascular endpoints, negative for cancer prevention.

Green coffee and chlorogenic acids

Chlorogenic acids are coffee's main polyphenol, and roasting degrades them — the entire rationale for green bean extract. A systematic review of three small RCTs (n=103) using at least 500 mg/day chlorogenic acid found a body weight reduction of 1.30 kg (95% CI −2.07 to −0.52), GRADE-downgraded for risk of bias and short duration. We make no weight-loss claims from that — green coffee's fame came from marketing that did not survive scrutiny. Grade: weak to moderate.

Turmeric and curcuminoids

Curcuminoids make up roughly 2–9% of turmeric by weight, and the story is bioavailability. Curcumin and its metabolites are undetectable in plasma below about 3.6 g/day; at that dose plasma runs near 0.01 µM, with rapid glucuronidation and sulfation dominating. Yet the mechanism is confirmed in humans: a randomized trial in 18 volunteers found oral curcumin induced NQO1 mRNA 3.3-fold in skin biopsies. Clinically, 0.5 g/day over nine months in prediabetes showed a large group difference in progression, while 4 g/day for six months in Alzheimer's disease was safe but produced no cognitive benefit. Piperine does raise plasma levels, partly by inhibiting glucuronidation and CYP enzymes — a drug-interaction mechanism, not a free lunch. Grade: strong for NRF2 engagement, moderate for inflammatory endpoints, negative for cognition.

Blueberry and cherry anthocyanins

A systematic review of 11 randomized trials in adults aged roughly 50–80 found benefits on memory, executive function, processing speed and brain perfusion — with the authors' own caveat that nine of the 11 were of poor or modest methodological quality. Promising and preliminary. Vascular effects rest on firmer ground: higher flavonoid intake is associated with roughly 10% lower cardiovascular risk and blood pressure reductions of 2–3 mmHg.

Tart cherry is the best-quantified of the seven for recovery. A meta-analysis of 19 trials and 385 participants found improved maximal voluntary contraction recovery (effect size 1.12 at 24 hours, 1.29 at 48 hours) and reduced CRP — but no significant effect on soreness, creatine kinase or IL-6, with heterogeneity of I² 69–93% and certainty very low to moderate. Dosing spans a 15-fold range in total polyphenols, so effect sizes do not transfer between products. The sleep trials have sample sizes in the single and low double digits: direction, not precision. Grade: moderate for functional recovery and CRP, null for soreness, preliminary for sleep.

Broccoli, sulforaphane, and the NRF2 story

Sulforaphane is one of the most potent naturally occurring NRF2 activators identified in food, and unusually it is long-acting: the isothiocyanate forms adducts with KEAP1 cysteines rather than transiently oxidizing them. In a randomized trial, 450 mg/day glucoraphanin induced NQO1 mRNA 3.1-fold in human skin biopsies (3.6-fold with curcumin), alongside reductions in IL-1β and TNF-α. Randomized broccoli sprout beverage trials in a heavily polluted region of China produced sustained, dose-dependent increases in urinary excretion of benzene and acrolein conjugates — a specific phase-II conjugation effect, and the reason to describe the chemistry rather than say "detox."

There is a real formulation catch. Glucoraphanin is inert until hydrolyzed by myrosinase, and cooking inactivates myrosinase. Without it, gut bacteria must do the job, poorly and variably — around 12% conversion on average, range 1.1% to 40.7%, against 35–41% recovery with active myrosinase. A broccoli ingredient without active or added myrosinase delivers a fraction of what its label implies. Grade: strong for enzyme induction, moderate for detoxification endpoints, unproven for disease outcomes.

Kale: kaempferol, quercetin, and honest limits

Kale is simultaneously a flavonol source (kaempferol and quercetin), a glucosinolate source, and one of the densest dietary sources of lutein and zeaxanthin. It is also the weakest of the seven for human intervention data — compositional analyses, animal feeding work and cell studies exist, but no human randomized trials of kale itself. Support is indirect: flavonol epidemiology, AREDS2 lutein and zeaxanthin data, and the broader cruciferous literature. Grade: weak/preliminary as a standalone ingredient.

A Note on VitAlign, the Complex We Use

Disclosure, in one labelled section. Current contains 50 mg of VitAlign® Cellular Health Complex — a plant-based polyphenol blend from those same seven foods, standardized to 60% polyphenols, 30% catechins and 3% curcuminoids. We chose a modest food-matrix dose because of the hormesis and inverted-U evidence above, not despite it.

Unusually for a botanical blend, it has been tested in humans at exactly that dose. Three peer-reviewed studies exist, and two of them are randomized and double-blind: Nemzer 2017 in Free Radical Research, on redox signalling; Nemzer 2021 in the Journal of Food Research, a double-blind, placebo-controlled, randomized longitudinal study of nitric oxide and mitochondrial metabolic activity; and Fink 2025 in the International Journal of Molecular Sciences, a randomized double-blind study that used vitamin C as an active comparator rather than a sugar pill. Between them they report reductions in markers of ROS activity across the NOX2, mitochondrial and iNOS pathways at 120 minutes, changes in mitochondrial function measures, and changes in circulating nitric oxide.

That is a stronger evidence base than most polyphenol blends can point to, and picking an active comparator over a placebo is a harder test than the category usually sets itself. Two caveats are still worth stating plainly. These are studies of the ingredient, not of Current — nobody has run a trial on the finished stick, and we would not imply otherwise. And the research is manufacturer-affiliated, which is entirely normal for a proprietary ingredient and is exactly why independent replication would strengthen the picture further.

When Not to Supplement Antioxidants

• If you smoke, do not take beta-carotene supplements. ATBC and CARET both found increased lung cancer in this exact population.
• Do not take gram-doses of vitamin C plus high-dose vitamin E around adaptation-focused training. The signals you are training to produce are the ones those doses suppress.
• Do not take high-dose vitamin E on anticoagulants without medical supervision. It interferes with vitamin K and raises bleeding risk.
• Be cautious with high-dose green tea extract, particularly fasted. EFSA's 800 mg EGCG/day transaminase threshold applies to supplements, not brewed tea.
• Do not stack selenium indiscriminately. RDA 55 µg/day, UL 400 µg/day. One ounce of Brazil nuts can supply over 500 µg.
• Do not take antioxidant supplements during chemotherapy or radiotherapy without your oncologist's approval. Some of those treatments work partly through oxidative mechanisms.
• Do not use antioxidants as a substitute for fruits and vegetables. The isolated high-dose trials are precisely the ones that failed.

The rule that falls out of all of it: correct shortfalls, support the enzyme systems, keep doses modest and food-shaped, and leave the signaling alone. That is why Current carries 50 mg of polyphenols and 500 mg of vitamin C rather than a gram of each, and why the trace mineral cofactors sit at levels that fill gaps rather than flood them. Life's better mineralized.

Where the Evidence Is Weak

• Whether polyphenol supplementation changes hard outcomes. Mechanism well characterized, biomarkers encouraging, long-term randomized outcome data barely exists.
• Whether stress, poor sleep and shift work genuinely raise oxidative stress in humans. Mechanistically compelling, biomarker evidence mixed, several studies frankly null.
• The optimal dose of anything here. The cocoa flavanol inverted U is one of very few human dose-response curves that exists.
• Whether biomarker changes matter clinically. Lowering an isoprostane is not the same as improving health.
• Individual variation in metabolism. Urolithin and equol production depend on your microbiome, and we cannot yet predict who produces them.
• Whether antioxidant intervention helps at altitude or against wildfire smoke. Exposure well documented. Intervention not.

Myths Worth Retiring

• "Free radicals are bad and antioxidants neutralize them." Wrong at first principles. Physiological ROS run insulin signaling, immune killing and training adaptation.
• "More antioxidants is better." Contradicted by ATBC, CARET, SELECT, Cochrane 2012, and by the cocoa flavanol inverted U.
• "High ORAC value means a healthier product." The USDA withdrew its own database over this exact misuse.
• "This supplement scavenges free radicals in your body." At 0.1–2 µM against 1–10 mM glutathione, arithmetically impossible.
• "Antioxidants reverse aging." The simple free-radical theory of aging did not survive contact with the data.
• "X has 20 times the antioxidants of blueberries." A tube-assay comparison with no in-vivo meaning.
• "Turmeric is highly absorbed." Curcumin is undetectable in plasma below roughly 3.6 g/day.
• "Antioxidants improve immunity by reducing oxidative stress." NOX2-derived ROS are the killing mechanism of neutrophils.
• "Beets, arginine and polyphenols all do the same thing for nitric oxide." Three distinct pathways with different evidence bases.

Frequently Asked Questions

What are antioxidants in simple terms?

Antioxidants are molecules that get oxidized so more important molecules — DNA, membrane lipids, enzymes — don't. Some are vitamins you eat, some are enzymes your body builds, and some, like polyphenols, are better described as signaling compounds that tell cells to build more of their own defenses.

What are free radicals and oxidative stress?

A free radical is a molecule with an unpaired electron, making it reactive. Not all reactive oxygen species are radicals — hydrogen peroxide and peroxynitrite are not. Oxidative stress is when production outruns your capacity to handle it, which redox biology distinguishes from eustress, the low-level signaling your cells depend on.

Should I take antioxidant supplements?

For most healthy, well-fed people, isolated high-dose antioxidant supplements have not shown disease-prevention benefit, and several large trials found harm. Modest food-derived polyphenol intake plus adequate selenium, zinc, copper and manganese is better supported than megadosing any single compound.

Who needs antioxidant support the most?

By evidence strength: older adults, people eating few fruits and vegetables, smokers, people with metabolic dysfunction, and people with high air pollution exposure. Athletes are nuanced — recovery support during congested schedules yes, chronic high-dose isolated vitamins around adaptation training no.

Can you take too many antioxidants?

Yes. The Cochrane review of 78 trials and 296,707 participants found a small but significant mortality increase in low-bias trials (RR 1.04). Beta-carotene increased lung cancer in smokers, vitamin E increased prostate cancer in SELECT, and cocoa flavanol benefits peaked then declined past roughly 710 mg.

Do antioxidants interfere with exercise?

High doses of isolated vitamin C and E can. Ristow's 2009 trial found they prevented the training-induced improvement in insulin sensitivity and blocked induction of PGC-1α and the body's own antioxidant enzymes. Paulsen's endurance study found attenuated mitochondrial adaptation markers; his strength study found no reduction in muscle gains.

What is an ORAC value, and does it mean anything?

ORAC measures how well a compound quenches a peroxyl radical in a test tube. The USDA withdrew its database in 2012, stating the values have no relevance to the effects of specific bioactive compounds and were routinely misused in marketing. It tells you nothing about what happens after absorption.

Are polyphenols antioxidants?

Technically yes, functionally mostly no. Plasma concentrations of 0.1–2 µM are three to four orders of magnitude below intracellular glutathione, so direct scavenging cannot be the main mechanism. They work by activating NRF2, which upregulates your own antioxidant enzymes.

Do antioxidants slow aging?

The simple free-radical theory of aging is not supported by current data. What is supported is that aging involves loss of redox signaling fidelity, declining NRF2 responsiveness and chronic low-grade inflammation. Dietary NRF2 activators address a mechanism implicated in that process, which is not the same as slowing aging.

Do minerals count as antioxidants?

No. Selenium, zinc, copper and manganese neutralize nothing themselves. They sit in the active sites of glutathione peroxidase and superoxide dismutase. Minerals don't fight free radicals — they build the machines that do, which is why we treat them as foundational in electrolytes versus minerals versus trace minerals.

Sources

1. Sies H, Jones DP. "ROS as pleiotropic physiological signalling agents." Nat Rev Mol Cell Biol, 2020. Link
2. Panday A et al. "NADPH oxidases: structure to innate immunity-associated pathologies." PMC4654378
3. ATBC Study Group. "Vitamin E and beta carotene on lung cancer in male smokers." NEJM, 1994. Link
4. Omenn GS et al. "Beta carotene and vitamin A on lung cancer" (CARET). NEJM, 1996. Link
5. Goodman GE et al. "CARET: six-year post-intervention follow-up." JNCI, 2004. Link
6. Klein EA et al. "Vitamin E and prostate cancer risk (SELECT)." JAMA, 2011. Link
7. Bjelakovic G et al. "Antioxidant supplements for prevention of mortality." Cochrane Database Syst Rev, 2012. Link
8. Ristow M et al. "Antioxidants prevent health-promoting effects of physical exercise." PNAS, 2009. Link
9. Paulsen G et al. "Vitamin C and E hamper cellular adaptation to endurance training." J Physiol, 2014. Link
10. Paulsen G et al. "Vitamin C and E alter protein signalling after strength training." J Physiol, 2014. Link
11. Fernando W et al. "Polyphenols generating H₂O₂ inhibit PTPs and activate the Nrf2 axis." Antioxidants, 2020. Link
12. Linus Pauling Institute. "Flavonoids." Link
13. Linus Pauling Institute. "Vitamin C." Link
14. Linus Pauling Institute. "Curcumin." Link
15. Linus Pauling Institute. "Tea." Link
16. Linus Pauling Institute. "Isothiocyanates." Link
17. NIH Office of Dietary Supplements. "Vitamin C." Link
18. NIH Office of Dietary Supplements. "Selenium." Link
19. "USDA/ARS ORAC database withdrawn." Nutraceuticals World, 2012. Link
20. Kissner R, Koppenol WH. "Superoxide with nitrogen monoxide: approaching the diffusion limit." J Phys Chem A, 2002. Link
21. Radi R. "The superoxide radical switch in the biology of nitric oxide and peroxynitrite." Physiol Rev, 2022. Link
22. "Dose–response between cocoa flavanols and human endothelial function." Food Funct, 2019. Link
23. CDC. "Adults meeting fruit and vegetable intake recommendations, US 2019." MMWR, 2022. Link
24. "Dietary polyphenols and all-cause mortality: meta-analysis." Metabolites, 2024. Link
25. EFSA. "Safety of green tea catechins." EFSA Journal, 2018. Link
26. Kanchanasurakit S et al. "Chlorogenic acid in green bean coffee on body weight." Syst Rev, 2023. Link
27. "Berry-based supplements and cognitive function." Sci Rep, 2022. Link
28. "Tart cherry juice and recovery from exercise-induced muscle damage." Sports Med Open, 2026. Link
29. Fahey JW et al. "Sulforaphane bioavailability: control by active endogenous myrosinase." PLoS ONE, 2015. Link
30. Egner PA et al. "Detoxication of airborne pollutants by broccoli sprout beverage." Cancer Prev Res, 2014. Link
31. "Oral glucoraphanin and curcumin modulate cytoprotective enzymes in human skin." Metabolites, 2025. Link
32. "Keap1-Nrf2 pathway in cancer." Front Oncol, 2024. Link
33. "Night shift work, oxidative stress and inflammation biomarkers." Environ Int, 2023. Link
34. "Variability of oxidative stress and inflammation biomarkers." Redox Rep, 2020. Link
35. "Urolithin A in highly trained male distance runners." 2025. PMC12628386
36. Harvard T.H. Chan School of Public Health. "Antioxidants." Link

These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease. This article is for educational purposes and is not medical advice. Talk to your doctor before starting any supplement, especially if you are pregnant, nursing, have kidney disease, or take prescription medication.