Basic research and case-control studies have suggested that citrus consumption may protect against cancer. However, the protective effect has been observed from few prospective studies. This study investigated the association of citrus consumption with cancer incidence among 42,470 Japanese adults in the Ohsaki National Health Insurance Cohort, which covered an age range of 40-79 years, and was followed up from 1995 to 2003 for all-cancer and individual cancer incidence. Citrus consumption was assessed using a self-administered questionnaire. The Cox proportional hazard model was applied to estimate relative risks (RRs) and 95% CIs. During the 323,204 person-years of follow-up, 3,398 cases were identified totally. Citrus consumption, especially daily consumption, was correlated with reduced all-cancer incidence, the RRs were 0.89 (95% CI = 0.80-0.98) for total participants, 0.86 (0.76-0.98) for males and 0.93 (0.79-1.09) for females, as well as multiple cancers at individual sites, especially pancreatic (RR = 0.62, 95% CI = 0.38-1.00) and prostate cancer (RR = 0.63, 95% CI = 0.41-0.97). Joint effect analysis showed a reduced risk of overall cancer existed only for subjects who consumed >or=1 cup green tea/day (RR = 0.83, 95% CI = 0.73-0.93) as well as for males (RR = 0.83, 95% CI = 0.71-0.97) or females (RR = 0.82, 95% CI = 0.68-0.99). These findings suggest that citrus consumption is associated with reduced all-cancer incidence, especially for subjects having simultaneously high green tea consumption. Further work on the specific citrus constituents is warranted, and clinical trials are ultimately necessary to confirm the protective effect.