The literature from 1990 to 2003 on the relation between coffee, decaffeinated coffee, tea and colorectal cancer risk has been reviewed. For the relation with coffee, three cohort (517 total cases) and nine case-control studies (7555 cases) analysed colon cancer; three cohort (307 cases) and four case-control studies (2704 cases) rectal cancer; six case-control studies (854 cases) colorectal cancer. For colon cancer most case-control studies found risk estimates below unity; the results are less clear for cohort studies. No relation emerged for rectal cancer. A meta-analysis, including five cohort and twelve case-control studies, reported a pooled relative risk of 0.76 (significant). Any methodological artefact is unlikely to account for the consistent inverse association in different countries and settings. Plausible biological explanations include coffee-related reductions of cholesterol, bile acids and neutral sterol secretion in the colon; antimutagenic properties of selected coffee components; increased colonic motility. Decaffeinated coffee was not related to either colon or rectal cancer in three case-control studies. No overall association between tea and either colon or rectal cancer risk emerged in seven cohort (1756 total cases of colon, 759 of rectal and 60 of colorectal cancer) and 12 case-control studies (8058 cases of colon, 4865 of rectal, 604 of colorectal cancer).