Tea is a drink, not a medicine. It can be delicious, stimulating and hydrating. It contains caffeine and a complicated mixture of plant compounds, some of which produce interesting effects in laboratory experiments and human studies. That does not make every claim about “antioxidants,” detoxification, immunity, fat burning or disease prevention true.

The most defensible conclusion is less dramatic: regular tea drinking is associated with some favourable health outcomes, particularly in cardiovascular research, and short trials report small changes in a few risk markers. Much of the long-term evidence is observational, however. It can show that tea drinkers and non-drinkers differed; it cannot by itself prove that tea caused the difference.

This article concerns tea made from Camellia sinensis—green, white, yellow, oolong, black and dark tea. It distinguishes brewed tea from concentrated green-tea extracts, because a cup, a bowl of matcha and a high-dose EGCG capsule are not interchangeable exposures.

The evidence at a glance

Common tea-health claims and the strength of the human evidence
Claim What the evidence supports Verdict
Tea improves alertness Caffeine can acutely improve alertness and attention; the amount in a cup varies widely. Supported, mainly because of caffeine
Tea is good for the heart Cohort studies often find lower cardiovascular and mortality risk among tea drinkers; short trials report small average changes in blood pressure or LDL cholesterol. Promising but not proven as disease prevention
Tea prevents dementia Several large cohorts find associations with lower risk, but no long-term randomised trial establishes prevention. Suggestive observational evidence
Green tea causes weight loss Controlled trials find little or no clinically meaningful effect. Not supported as a useful weight-loss method
Tea prevents cancer Laboratory findings are abundant; results in humans are inconsistent. Unproven
Tea prevents flu or “boosts immunity” Small studies of catechin products and tea gargling are interesting but do not establish that ordinary tea prevents infection. Preliminary and insufficient
Tea detoxifies the body No clinical meaning, toxin or measurable outcome is usually specified, and no convincing evidence supports the claim. Marketing language, not an established effect
Tea dehydrates you Ordinary tea contributes to fluid intake; its water is not cancelled out by its caffeine. False at normal drinking levels

Why tea-health headlines become more certain than the science

“Antioxidant” describes chemistry, not a medical outcome

Tea contains polyphenols. Green tea is rich in catechins such as EGCG, while oxidation during black-tea production creates theaflavins and other compounds. These substances can participate in oxidation-reduction reactions, and researchers investigate direct antioxidant activity as well as effects on cell signalling and enzymes.

A test-tube result does not show that a cup protects a person from cancer, dementia or ageing. Compounds are absorbed, transformed by digestion and the microbiome, circulated at much lower concentrations and eventually excreted. A change in a laboratory antioxidant assay is not the same endpoint as fewer heart attacks or longer life.

This distinction is not pedantry. The European Food Safety Authority’s assessment of tea and oxidative-damage claims did not establish the required cause-and-effect relationship. Tea unquestionably contains polyphenols; “therefore it protects your cells” is the unsupported extra step.

Association is not causation

Researchers cannot realistically assign thousands of people to drink or avoid tea for several decades while controlling the rest of their lives. Long-term tea research therefore relies heavily on cohort studies: participants report what they consume, and researchers compare later diagnoses or deaths.

Those studies can adjust statistically for smoking, income, diet, exercise and other measured factors, but never for everything. Tea drinking may travel with other habits. Illness can also change what a person drinks, creating reverse causation. Large samples make an estimate more precise; they do not automatically remove these problems.

A relative percentage is not an individual promise

“Twenty per cent lower risk” usually describes a relative difference between groups. If the underlying risk were 5%, a 20% relative reduction would mean 4% rather than 5%—an absolute difference of one percentage point. That example is not a tea-study result; it shows why a headline percentage is incomplete without the baseline risk, follow-up time and study design.

“Cup” is imprecise too. A Japanese teacup, a British mug and a large takeaway drink can differ several-fold in volume. Leaf quantity, cultivar, brewing time, repeated infusions, milk, sugar and whether the study concerned green tea or black tea add still more variation.

Brewed tea is not a catechin capsule

Trials often use a standardised extract because its dose is easier to control. That can answer a question about the extract, but not necessarily about normal tea drinking. Conversely, a reassuring history of drinking tea cannot establish the safety of swallowing a concentrated extract all at once. Any article that moves casually between cups, matcha, isolated EGCG and supplements is mixing different interventions.

What human research actually says

Alertness and short-term cognitive performance

The clearest immediate effect of ordinary tea comes from caffeine. It blocks adenosine receptors and can increase alertness and attention. It can also disturb sleep or cause restlessness, nausea, tremor and palpitations in sensitive people. Tea often contains less caffeine per serving than coffee, but “tea” is not one fixed dose.

L-theanine is also studied for possible effects on attention, stress and mood, alone and with caffeine. Results are interesting but variable, and many experiments use isolated doses unlike those in an ordinary cup. L-theanine does not bind caffeine or turn tea into a slow-release stimulant. Feeling calmer with tea is a valid personal experience; it is not proof of that popular mechanism.

Heart health and longevity

This is where the case for a possible benefit is strongest—and where careful wording still matters. In a UK Biobank cohort of almost half a million participants, people reporting at least two cups a day had a 9–13% lower relative risk of death than non-drinkers during follow-up. The study was large and adjusted for many variables, but it remained observational. The US National Institutes of Health described the result as an association, not proof that two cups prolong life.

Asian cohorts have reported similar patterns in populations with different tea habits. The China-PAR project, for example, associated consistent habitual tea drinking with lower cardiovascular and all-cause mortality risk. That cross-cultural consistency is worth investigating, but several cohorts agreeing with one another do not become a randomised trial.

Shorter controlled trials provide a different piece of the puzzle. Reviews report small average reductions in blood pressure or LDL cholesterol in some comparisons, while other analyses find no significant effect. Trials are often short, small and based on extracts rather than a familiar pot of tea. A two-millimetre change in blood pressure is a biomarker result, not direct evidence that tea prevents strokes or heart attacks.

Verdict: the repeated observational signal is plausible and deserves further research. It should not be sold as a treatment, a substitute for prescribed medicine or a guaranteed route to longevity.

Dementia and long-term brain health

The dramatic dementia percentages also come from observational research. A 2022 UK Biobank study associated moderate tea intake with lower rates of dementia during roughly nine years of follow-up. A newer Japanese cohort associated at least 600 ml of green tea a day with about 25% lower relative dementia risk than the lowest-intake group.

Neither study randomly assigned tea. Intake was self-reported, drinkers differed in other ways, cup sizes and tea chemistry varied, and dementia develops over many years. It is therefore inaccurate to convert the statistical model into “every extra cup gives 5% more protection.” A dose-response association inside one cohort remains an association.

Verdict: the evidence justifies further research, not a preventive dose. Caffeine can make a person more alert today; that is a separate claim from preventing neurodegenerative disease decades later.

Weight loss and metabolism

Catechins and caffeine can affect energy expenditure in short experiments, which gave the “fat-burning tea” claim a plausible mechanism. The clinically useful question is whether people actually lose a meaningful amount of weight.

A Cochrane review of controlled trials concluded that green-tea preparations produced a very small, statistically non-significant change outside Japan and were unlikely to have a clinically important effect. They did not meaningfully help maintain weight loss.

Unsweetened tea can still be useful if it replaces a high-sugar drink: the practical difference then comes mainly from what is no longer in the cup. Adding sugar, syrup, sweetened condensed milk or a large portion of tapioca naturally changes that comparison.

Verdict: enjoy tea during weight loss if you like it, but do not expect tea itself to cause meaningful fat loss.

Cancer prevention

Tea compounds can affect cancer cells and tumour pathways in laboratory and animal models. Many substances do. The difficulties are achieving relevant exposure safely in humans and showing that people who consume tea actually develop fewer cancers because of it.

Human studies vary by cancer site, population, tea type and study design, and positive findings are not consistent. The US National Center for Complementary and Integrative Health concludes that current evidence cannot establish whether green tea prevents cancer. Tea should never be presented as cancer treatment.

There is also an important reversal of the usual claim: regularly drinking any beverage at temperatures above about 65 °C is classified by the International Agency for Research on Cancer as probably carcinogenic to humans, based on evidence concerning oesophageal cancer. The issue is repeated thermal injury, not tea as an ingredient. Let a painfully hot drink cool.

Verdict: tea has not been shown to prevent or treat cancer. Avoiding scalding drinking temperatures is better supported than claiming protection from catechins.

Type 2 diabetes and blood-sugar control

Some cohorts associate higher tea consumption with lower rates of type 2 diabetes, but pooled results are inconsistent and apparent benefits often appear only in higher-intake subgroups. Trials of green tea or extracts report mixed results for fasting glucose, insulin and HbA1c. These studies do not establish tea as prevention or treatment for diabetes.

The drink’s formulation may matter more than a subtle catechin effect. Plain tea contains practically no sugar; bubble tea, bottled iced tea and heavily sweetened chai can contain a substantial amount. Calling all of them “tea” does not give them the same metabolic profile.

Verdict: choose unsweetened tea if reducing added sugar is the goal, but do not use it in place of evidence-based diabetes care.

Immunity, colds and influenza

The often-repeated “about 33% less influenza” figure needs its missing context. A 2021 systematic review and meta-analysis combined small Japanese and other studies of several different interventions: tea gargling, catechin-containing drinks and catechin capsules. Its pooled estimates favoured the interventions, but the authors identified possible publication bias, most studies were conducted in Japan, and only some examined oral consumption. The review was also funded by Kao Corporation, a producer of high-catechin green-tea beverages; funding does not invalidate a result, but it belongs in a transparent reading.

That evidence is too indirect and heterogeneous to say that ordinary green tea reduces a person’s flu risk by 33%. “Supports immunity” is vaguer still: it rarely identifies which immune measurement should change or whether people actually become ill less often.

Verdict: tea may be a pleasant drink when you are well or unwell, but it is not an established method of preventing influenza and does not replace vaccination or medical care.

Real effects and sensible cautions

Tea contributes to hydration

Caffeine can have a mild diuretic effect, especially in larger doses or in people unaccustomed to it. That does not make the water in tea disappear. In a controlled crossover trial, four or six mugs of black tea produced hydration measures comparable with water in healthy men. Ordinary tea counts towards fluid intake. Read the hydration trial.

Tea can reduce non-haem iron absorption

Tea polyphenols can bind non-haem iron during digestion, so drinking tea with a meal can reduce absorption of the iron found especially in plant foods and fortified products. This does not mean that every vegetarian must give up tea. It matters most for people with low iron stores, diagnosed iron-deficiency anaemia, high requirements or diets relying heavily on non-haem iron.

A controlled human study found that waiting one hour after an iron-containing meal attenuated the effect. People being treated for iron deficiency should follow the timing advice given for their meals and supplements rather than relying on a general website rule. Read the iron-absorption study.

Caffeine tolerance is individual

The European Food Safety Authority reports that single doses up to 200 mg and total intake up to 400 mg a day do not raise safety concerns for most healthy adults. These are upper safety assessments, not recommended targets. EFSA also notes that 100 mg near bedtime can affect sleep in some adults and gives a lower 200 mg daily level for pregnancy.

Caffeine varies with leaf, dose and preparation. A light cup, a long-steeped pot and matcha can differ greatly. If tea causes insomnia, anxiety, palpitations, nausea or headaches, your useful limit is lower than a population-level ceiling.

Concentrated green-tea extracts carry different risks

Brewed green tea is generally considered safe for adults. Concentrated extracts are the clearer concern: reported liver injuries have predominantly involved extract products. EFSA found that supplemental doses at or above 800 mg EGCG per day may be associated with initial signs of liver injury and could not identify a universally safe supplemental dose below that threshold. EFSA’s safety summary distinguishes infusions from supplements.

EU rules now restrict specified foods containing added green-tea extracts, require warnings including not to consume them on an empty stomach, and prohibit a daily portion containing 800 mg or more of EGCG. Those rules do not turn brewed tea into a dangerous drink; they show why “more concentrated” is not the same as “more beneficial.” See Commission Regulation (EU) 2022/2340.

Tea and medicines can interact

The best-documented interactions often concern concentrated green-tea products or unusually high intake. Green tea can reduce exposure to some medicines, including the beta-blocker nadolol, and extracts may interact with other drugs. Anyone taking medicine with food-and-drink instructions, cancer treatment, anticoagulants or iron therapy should ask a pharmacist or clinician about that specific medicine rather than assuming that “natural” means irrelevant. The NCCIH green-tea safety overview lists known cautions and interactions.

So, is tea healthy?

Tea can fit into many dietary patterns. Plain tea provides fluid and sensory pleasure with little or no energy, and its caffeine can improve alertness. Regular drinkers in many large cohorts also show favourable cardiovascular and mortality associations. The associations are worth studying, but they are not personal guarantees.

The evidence does not support turning tea into preventive medicine. It does not detoxify the body, melt fat, guarantee a longer life, prevent cancer or provide a measured percentage of protection from dementia or influenza. A good cup does not need imaginary powers to deserve its place in daily life.

Questions about tea and health

How many cups should I drink for health benefits?

There is no clinically established tea dose and no universal recommendation to drink two, three or four cups. Cohort-study categories are not prescriptions. Drink an amount you enjoy and tolerate while considering total caffeine, sleep, pregnancy, iron status and any relevant medical advice.

Which type of tea is healthiest?

No tea category has been shown to be the universal healthiest choice. Green, white, yellow, oolong, black and dark teas contain different mixtures of compounds, and variation within each category can be considerable. Choose by flavour and preparation rather than treating the six tea types as a medical ranking.

Is tea healthier than coffee?

The evidence cannot award a simple winner. Both drinks are associated with some favourable outcomes in observational research, both contain biologically active compounds, and both can disturb sleep or cause caffeine symptoms. Tea often—but not always—provides less caffeine per serving. The better choice is the one you enjoy, tolerate and do not load with unwanted sugar.

Can I take green-tea supplements instead?

They are not a concentrated shortcut to the unproven benefits of tea. Extracts expose the body to different doses and can cause adverse effects, including rare liver injury. Do not assume that a supplement is safer or more effective because its ingredient originated in tea.

Should vegetarians avoid tea?

Not routinely. Tea principally affects non-haem iron absorption when consumed around the same time as the meal. People at risk of deficiency can separate tea from iron-rich meals and discuss their individual needs with a qualified professional; people with no iron problem do not need to fear every cup.

Does adding milk destroy tea’s benefits?

There is no proven package of benefits for milk to “destroy.” Studies of milk and tea polyphenol absorption or vascular markers have produced mixed results, and ordinary use of milk is not a reason to declare a cup worthless. Sugar and portion size are separate questions.

Further reading at RealiTea

Evidence and sources