Polyphenols: what tea, fruit and supplements actually do

Polyphenols are sold as a broad antioxidant shield, yet the benefit results with numbers come from concentrated supplements, often given to patients, such as a 2-point gain on a cognition test across 13 trials of 849 people with neurodegenerative diseases. For everyday eating the evidence is intake surveys and observational links. The clearest dose findings concern green tea extract capsules, and a supplement mix that cut iron absorption in people with iron overload.

How much you need

• Australians aged 2 and over eat a median of about 480 mg of polyphenols a day, mostly flavonoids, with tea and coffee the top source. — 7,735 Australians aged 2+, national survey 2011-12 Evidence C source
• Chilean adults reported about 1,635 mg of polyphenols a day, with fruit as the main source followed by coffee and tea. — 521 adults in Chile Evidence C source
• In 22,169 Italian adults, the highest ultra-processed food intake went with about 132 mg a day less dietary polyphenol than the lowest. — 22,169 men and women, Moli-sani Study Evidence C source

How well your body absorbs it

• In a crossover trial in 14 people with hereditary hemochromatosis, a supplement of black tea, cocoa and grape extract taken with a meal cut non-heme iron absorption by about 40%. — 14 patients with HFE p.C282Y hereditary hemochromatosis Evidence B source
• In a double-blind crossover RCT in 40 patients with iron overload, 100 mg of procyanidins with an iron-rich meal did not reduce iron absorption. — 20 hereditary hemochromatosis and 20 dysmetabolic iron overload patients Evidence B source

Who is most likely to fall short

• A meta-analysis of 14 trials in pregnancy found only low-certainty single-trial signals for EGCG or resveratrol added to nifedipine in preeclampsia, and evidence on benefits and harms remains limited. — pregnant women, 14 RCTs; EGCG and resveratrol signals from 1 trial of 349 women each Evidence A source

What the evidence does and does not show

• In 13 RCTs of 849 people with neurodegenerative diseases, polyphenol supplements raised MMSE cognition scores by about 2 points. — 849 participants with seven neurodegenerative diseases, 13 RCTs Evidence A source
• A meta-analysis of 21 RCTs with 1,502 participants found flavonoid and anthocyanin supplements lowered serum uric acid, while stilbenes and curcuminoids did not. — 21 RCTs, 1,502 participants Evidence A source
• Pooling 4 human intervention studies, polyphenol intake raised gut Bifidobacterium abundance with a moderate-to-large effect (SMD 0.81). — 22 human intervention studies reviewed, 4 pooled Evidence B source
• Observational data from 10 studies of 106,302 people link the highest polyphenol intakes to 9% lower odds of obesity. — 10 studies, 106,302 participants Evidence C source
• The only polyphenol health claim authorized in the EU snapshot says olive oil polyphenols help protect blood lipids from oxidative stress, while 18 other polyphenol claims were rejected. — EU claims register, claim POL-HC-6431 Evidence E source

What too much does

• Across 34 RCTs of green tea extracts, liver-related adverse events were rare and mild, with a non-significant odds ratio of 2.1 versus placebo. — 34 RCTs of green tea extracts Evidence A source
• In the 12-month Minnesota Green Tea Trial, 843 mg a day of EGCG raised ALT by about 78% at 6 months in women with the UGT1A4 A/C genotype versus 28% in C/C carriers. — postmenopausal women, 400 in the green tea extract arm, 12 months Evidence B source
• Laboratory data and 159 human studies suggest green tea liver toxicity depends on dose form, and the review derived a safe level of 338 mg EGCG a day for bolus doses. — toxicology studies and 159 human intervention studies Evidence D source

Medicines it interacts with

• In a randomized study in healthy volunteers, pomegranate juice and a 1 g pomegranate extract did not change clearance of the CYP2C9 drug flurbiprofen, despite inhibiting the enzyme in the lab. — healthy human volunteers Evidence B source

What to do with this

Most of the polyphenol in an ordinary diet comes from drinks and fruit. Australians aged 2 and over ate a median of about 480 mg a day, mostly flavonoids, with tea and coffee the top source. Chilean adults reported about 1,635 mg a day, with fruit first, on a questionnaire method that tends to overestimate.

Among 22,169 Italian adults, those eating the most ultra-processed food took in about 132 mg a day less polyphenol than those eating the least. That comes from a single cross-sectional survey.

Green tea extract is where dose matters. Across 34 trials of green tea extracts, liver problems were rare and mild. In a 12-month trial of 843 mg of EGCG a day in capsules, the liver enzyme ALT rose 78 percent by month 6 in postmenopausal women with the UGT1A4 A/C genotype, against 28 percent in C/C carriers. One review derived 338 mg of EGCG a day as a safe level for solid bolus doses, a figure set by its authors and not by a regulator. If you buy a green tea extract, the EGCG per capsule on the label is the number to compare.

If you have hereditary hemochromatosis, note that a supplement of black tea, cocoa and grape extract taken with a meal cut non-heme iron absorption by about 40 percent in a crossover trial of 14 people. A single 100 mg dose of procyanidins did not reduce absorption in 40 patients with iron overload. How tea and cocoa fit around your meals is a question for the clinician managing your iron.

In pregnancy, 14 trials gave only single-trial, low-certainty signals for EGCG or resveratrol added to nifedipine in preeclampsia. Polyphenol supplements in pregnancy are a question for whoever is managing it.

Where to get it from food

USDA does not report polyphenols for individual foods, so this site cannot rank foods by it. The sections above name the foods the studies themselves looked at.

The bottom line

The polyphenol benefit results with real numbers come from supplements, often in people with a disease, and they say nothing direct about a cup of tea or a bowl of berries in a healthy person. Observational data link the highest intakes to 9 percent lower odds of obesity, an association with high heterogeneity. The dose findings that hold up concern concentrated green tea extract, which raised a liver enzyme in some women at 843 mg of EGCG a day, and iron absorption in people who already carry too much iron. Apart from green tea extract, we have no card on an upper limit for polyphenol supplements.

Sources

1. plyr3-12 · Observational data · cross-sectional survey
   The estimated total polyphenol intake of Australians aged 2+ years [mean (SD) age: 40.9 (22.7) years] was median (IQR) 479.07 (529.81) mg/day when unadjusted for energy intake
   J Hum Nutr Diet, 2026 · checked 2026-10-02 · independently re-checked
2. plyr3-13 · Observational data · cross-sectional
   The average daily intake of polyphenols was 1634.9 ± 1168.6 mg/d
   J Am Nutr Assoc, 2026 · checked 2026-10-02 · independently re-checked
3. plyr3-14 · Observational data · cross-sectional
   Higher UPF intake was associated with lower dietary polyphenol content (β = -132.3; 95 % CI: -139.3, -125.4 mg/d), compared to the bottom category
   Nutr Metab Cardiovasc Dis, 2026 · checked 2026-10-02 · independently re-checked
4. plyr3-06 · Randomized controlled trial(s) · RCT
   In the iron absorption study, the PPS reduced FIA by ~ 40%: FIA from the meal consumed with the PPS was lower (3.01% (1.60, 5.64)) than with placebo (5.21% (3.92, 6.92)) (p = 0.026))
   Eur J Nutr, 2022 · checked 2026-10-02 · independently re-checked
5. plyr3-07 · Randomized controlled trial(s) · RCT
   20 HH and 20 DIOS patients were enrolled in a double-blind three-period crossover randomized study. ... Procyanidin supplementation does not significantly reduce dietary iron absorption in either disease.
   Clin Nutr, 2020 · checked 2026-10-02 · independently re-checked
6. plyr3-05 · Meta-analysis or systematic review · meta-analysis of RCTs
   the addition of epigallocatechin gallate (EGCG) to nifedipine may reduce the time needed to achieve blood pressure control (mean difference (MD) = -14.10 min, 95% CI -18.46 to -9.74) ... evidence on the clinical and adverse effects of polyphenols is limited and uncertain.
   BJOG, 2025 · checked 2026-10-02 · independently re-checked
7. plyr3-01 · Meta-analysis or systematic review · meta-analysis of RCTs
   13 studies involving 849 participants were included. ... Overall, polyphenol supplementation was associated with improved global cognition (pooled MD in MMSE = 2.06; 95% CI 0.62-3.49).
   Food Funct, 2026 · checked 2026-10-02 · independently re-checked
8. plyr3-02 · Meta-analysis or systematic review · meta-analysis of RCTs
   A total of 21 RCTs with 1502 participants were included. ... polyphenols lementation, particularly flavonoids and anthocyanin mixtures, significantly reduced SUA levels, while stilbenes and curcuminoids showed non-significant effects with current evidence.
   Crit Rev Food Sci Nutr, 2026 · checked 2026-10-02 · independently re-checked
9. plyr3-03 · Randomized controlled trial(s) · meta-analysis
   The meta-analysis revealed a significant positive effect of polyphenol intake on Bifidobacterium abundance (SMD = 0.81; 95% CI: 0.18-1.44; p = 0.0114), corresponding to a moderate-to-large effect size.
   Nutrients, 2026 · checked 2026-10-02 · independently re-checked
10. plyr3-04 · Observational data · meta-analysis of observational studies
    Finally, 10 studies with 106,302 participants were included in the meta-analysis. When different polyphenolic compounds were pooled, they were linked to lower odds of obesity (OR: 0.91; 95%CI: 0.85–0.97)
    BMC Endocr Disord, 2026 · checked 2026-10-02 · independently re-checked
11. plyr3-15 · Position of an expert body · regulatory decision
    POL-HC-6431 Olive oil polyphenols Olive oil polyphenols contribute to the protection of blood lipids from oxidative stress Authorised (value from the source's table, not a sentence)
    EU Register on nutrition and health claims, claim POL-HC-6431, snapshot 2026-09-21 · checked 2026-10-02 · independently re-checked
12. plyr3-08 · Meta-analysis or systematic review · systematic review of RCTs
    The summary odds ratio, estimated using a meta-analysis method for sparse event data, for intervention compared with placebo was 2.1 (95% confidence interval: 0.5-9.8). The few events reported in both groups were elevations of liver enzymes. Most were mild, and no serious liver-related adverse events were reported.
    Eur J Clin Nutr, 2016 · checked 2026-10-02 · independently re-checked
13. plyr3-09 · Randomized controlled trial(s) · RCT
    participants were randomized to consume high-dose GTE (843 mg/day EGCG) or placebo capsules for 12 months. ... From baseline to months 6 and 9, respectively, mean ALT change was +78.1% and +82.1% in the UGT1A4 (rs6755571) A/C genotype versus +28.0% and +30.1% in the C/C genotype
    J Diet Suppl, 2023 · checked 2026-10-02 · independently re-checked
14. plyr3-10 · Mechanistic or laboratory data · systematic review
    A safe intake level of 338 mg EGCG/day for adults was derived from toxicological and human safety data for tea preparations ingested as a solid bolus dose. An Observed Safe Level (OSL) of 704 mg EGCG/day might be considered for tea preparations in beverage form based on human AE data.
    Regul Toxicol Pharmacol, 2018 · checked 2026-10-02 · independently re-checked
15. plyr3-11 · Randomized controlled trial(s) · RCT
    In clinical studies, flurbiprofen pharmacokinetics were unchanged by pomegranate juice or extract as compared to a low-polyphenol placebo control beverage.
    Clin Pharmacol Ther, 2012 · checked 2026-10-02 · independently re-checked