Although long-chain n-3 polyunsaturated fatty acids (Ln-3 PUFA), which are abundant in fish, have shown protective effects on colorectal cancer in laboratory studies, epidemiological studies to date have not been consistent. We evaluated the relationship of consumption of fish and Ln-3 PUFA to the colon and rectal cancer risk in the two cohorts of the Japan Public Health Center-based prospective study of 42,525 men and 46,133 women. Dietary and other exposure data were obtained between 1990 and 1994. Through December 1999, 705 cases of colon and rectal cancer were documented. When data from the two cohorts were pooled, multivariable relative risks (RRs) for the highest quartile compared with the lowest quartile of fish consumption were 1.07 (95% confidence interval, CI = 0.77-1.48) for colon cancer and 0.95 (95% CI = 0.63-1.43) for rectal cancer with no dose-risk trend. RRs for the highest quartile compared with the lowest quartile of eicosapentaenoic acid consumption were 1.05 (95% CI = 0.76-1.46) for colon cancer and 0.91 (95% CI = 0.60-1.38) for rectal cancer with no dose-risk trend. This study does not support the role of fish and Ln-3 PUFA in the etiology of colon and rectal cancer in this population whose fish consumption was high and the variation in Ln-3 PUFA consumption was large.