OBJECTIVE Laboratory and animal studies have shown a protective effect of green tea on cancer of different sites, but epidemiological evidence is limited and inconclusive. This prospective study in Japan examined the association between green tea consumption and cancer incidence. METHODS Subjects were 38,540 people (14,873 men, mean age 52.8 years; 23,667 women, mean age 56.8 years) who responded to a mail survey carried out between 1979 and 1981. A self-administered questionnaire ascertained consumption frequency of green tea using precoded answers (never, once per day, twice to four times per day, and five or more times per day). Follow-up continued until 31 December 1994. The study analyzed solid cancers (n = 3881); hematopoietic cancers (188); cancers of all sites combined (4069); and cancer of specific sites with more than 100 cases, i.e. stomach (901), colon (432), rectum (193), liver (418), gallbladder (122), pancreas (122), lung (436), breast (281), and bladder (122). Poisson regression was used to allow for city, gender, age, radiation exposure, smoking status, alcohol drinking, body-mass index, education level, and calendar time. RESULTS Green tea consumption was virtually unrelated to incidence of cancers under study. The relative risks of all cancers for those consuming green tea twice to four times per day and five or more times per day were 1.0 (95% confidence interval 0.91-1.1) and 0.98 (0.88-1.1), respectively, as compared with those consuming green tea once per day or less. CONCLUSIONS Our findings do not provide evidence that regular green tea consumption is related to reduced cancer risks.