What is actually in the cup
Tea from Camellia sinensis contains polyphenols — chiefly catechins, of which epigallocatechin-3-gallate (EGCG) is the most studied — plus caffeine and L-theanine. Herbal infusions are a different world: chamomile brings apigenin, peppermint menthol, ginger gingerols.
Real compounds, real pharmacology. But the distance between "this molecule does something" and "drinking this will help you" is where nearly all the confusion lives.
Step one: a compound does something in a dish
Most herbal research starts in vitro — cells in a plate, exposed to a purified compound. It is cheap, fast, and the source of most exciting headlines.
It is also the weakest possible evidence about a person. The concentrations used are often far higher than a human body ever achieves, and a cell in a plate has no stomach, no gut bacteria, and no liver.
Step two: the bioavailability problem
Human pharmacokinetics makes the gap concrete. When eight volunteers drank green tea and had their blood sampled repeatedly, peak plasma EGCG reached roughly 78 nanograms per milliliter, with a half-life near 3.4 hours and wide variation between people — and between repeat sessions in the same person. Epigallocatechin and epicatechin circulated mostly in conjugated forms — altered by the body, no longer the molecule the petri dish received [1].
So when a paper says EGCG kills cancer cells at some concentration, ask whether your bloodstream ever comes near it after a mug of sencha. Usually it does not.
Step three: animals
Animal studies add a whole organism, with digestion, metabolism, and excretion. That is a real advance, and still a poor predictor of human outcomes: doses are often far above realistic intake, rodent metabolism differs from ours, and the condition modeled is rarely the human disease. Many herbal claims have never left this stage.
Step four: humans, in small numbers
A randomized controlled trial in people is the first evidence that means much. Most herbal RCTs are small, short, and run exactly once.
A frequently cited trial of chamomile extract for generalized anxiety disorder randomized 57 outpatients over eight weeks and found a modest reduction in anxiety scores that just cleared statistical significance. The authors called it the first controlled trial of its kind and asked for replication [2]. Promising, not settled — and participants swallowed a standardized capsule, not a teabag.
Step five: pooling the trials, and disagreeing about them
Meta-analyses combine trials and sit at the top of the hierarchy. They still disagree with each other.
A 2019 meta-analysis pooled twelve randomized trials of peppermint oil in 835 people with irritable bowel syndrome and reported a clear benefit for global symptoms [3]. An umbrella review later re-examined nutritional interventions for IBS using GRADE — a formal system for grading how much confidence a body of evidence deserves — and rated most of those same effects as low or very low certainty [4]. Same trials, different question: not "is there an effect?" but "how much should we trust the answer?"
Cochrane's review of green tea and cancer prevention makes the point at scale: 142 studies, eleven of them randomized trials, plus observational data on more than a million people, yielding inconsistent results and limited evidence [5].
Dose and form change everything
A concentrated extract is not a beverage. A United States Pharmacopeia expert review found that green tea extract taken as a fasting bolus dose substantially increases EGCG absorption, and that published case reports of liver injury clustered at EGCG intakes from about 140 mg to roughly 1,000 mg per day [6]. Herbal and dietary supplements now account for about a fifth of cases in the US Drug-Induced Liver Injury Network [7]. We never blur brewed tea with a capsule labeled "tea extract."
How to read our labels
- Strong — multiple randomized controlled trials in humans, or a meta-analysis of them, pointing the same way. Our wording: "helps," "reduces."
- Moderate — at least one reasonably well-conducted human RCT, or consistent human observational data. Our wording: "may help."
- Limited — small or low-quality human trials, or mostly test-tube and animal work. Our wording: "one small trial found," "early research suggests."
- Traditional — long-standing traditional or historical use with little or no human clinical evidence. Our wording: "traditionally used for." This is the most common label on the site.
Why "traditional" is not an insult
Traditional use is where medicine has always found its leads; aspirin and digoxin both began as folk remedies. It tells you a plant has been consumed by many people for a long time, which is weak but genuine safety information. It does not tell you the plant works. Both are true at once.
References
The citations below were retrieved from PubMed.
- Lee MJ, Maliakal P, Chen L, et al. Pharmacokinetics of tea catechins after ingestion of green tea and (-)-epigallocatechin-3-gallate by humans: formation of different metabolites and individual variability. Cancer Epidemiol Biomarkers Prev. 2002;11(10 Pt 1):1025–1032. PMID: 12376503
- Amsterdam JD, Li Y, Soeller I, et al. A randomized, double-blind, placebo-controlled trial of oral Matricaria recutita (chamomile) extract therapy for generalized anxiety disorder. J Clin Psychopharmacol. 2009;29(4):378–382. doi:10.1097/JCP.0b013e3181ac935c (PMID: 19593179)
- Alammar N, Wang L, Saberi B, et al. The impact of peppermint oil on the irritable bowel syndrome: a meta-analysis of the pooled clinical data. BMC Complement Altern Med. 2019;19(1):21. doi:10.1186/s12906-018-2409-0 (PMID: 30654773)
- Zeraattalab-Motlagh S, Ranjbar M, Mohammadi H, Adibi P. Nutritional interventions in adult patients with irritable bowel syndrome: an umbrella review of systematic reviews and meta-analyses of randomized clinical trials. Nutr Rev. 2025;83(3):e1343–e1354. doi:10.1093/nutrit/nuae107 (PMID: 39110917)
- Filippini T, Malavolti M, Borrelli F, et al. Green tea (Camellia sinensis) for the prevention of cancer. Cochrane Database Syst Rev. 2020;3(3):CD005004. doi:10.1002/14651858.CD005004.pub3 (PMID: 32118296)
- Oketch-Rabah HA, Roe AL, Rider CV, et al. United States Pharmacopeia (USP) comprehensive review of the hepatotoxicity of green tea extracts. Toxicol Rep. 2020;7:386–402. doi:10.1016/j.toxrep.2020.02.008 (PMID: 32140423)
- Halegoua-DeMarzio D, Navarro V. Challenges in herbal-induced liver injury identification and prevention. Liver Int. 2025;45(3):e16071 (epub 2024). doi:10.1111/liv.16071 (PMID: 39136211)
Last reviewed and updated . HelperTea is written by an enthusiast, not a clinician, and is not medically reviewed. How we research and rate evidence. Found an error? Tell us — safety corrections get priority.