Several clinical studies have suggested that supplementation with fish oils can suppress the proliferation of colorectal mucosa and therefore inhibit the development of colorectal cancer. However, epidemiological evidence concerning fish consumption and risk is inconsistent and limited. To clarify the association between intake of fish and the likelihood of developing colorectal cancer, we conducted a large sample size case–reference study with 928 cases of colon cancer, 622 of rectal cancer and 46 886 cancer-free outpatient references aged 40–79 years. The data showed frequent raw/cooked fish intake to be associated with decreased odds ratio (OR) 0.68 with 95% confidence interval (CI) 0.47–0.99 for male colon cancer, especially for males aged over 60 years, smokers and frequent meat eaters. A marginal decrease in the OR (OR 0.58, 95% CI 0.31–1.07) was also detected for female rectal cancer, especially in the regular physical exercise subgroup. However, frequent dried/salted fish intake was found to be associated with increased OR in females younger than 60 years old and alcohol drinkers. Although there is some possible bias in epidemiological studies, the results suggest that frequent raw/cooked fish intake may decrease the risk while dried/salted fish, in contrast, may exert a detrimental effect.