Dietary intake of cruciferous vegetables (Brassica spp.) has been inversely related to colorectal cancer risk, and this has been attributed to their high content of glucosinolate degradation products such as isothiocyanates (ITCs). These compounds act as anticarcinogens by inducing phase II conjugating enzymes, in particular glutathione S-transferases (GSTs). These enzymes also metabolize ITCs, such that the protective effect of cruciferous vegetables may predicate on GST genotype. The Singapore Chinese Health Study is a prospective investigation among 63 257 middle-aged men and women, who were enrolled between April 1993 and December 1998. In this nested case-control analysis, we compared 213 incident cases of colorectal cancer with 1194 controls. Information on dietary ITC intake from cruciferous vegetables, collected at recruitment via a semi-quantitative food frequency questionnaire, was combined with GSTM1, T1 and P1 genotype from peripheral blood lymphocytes or buccal mucosa. When categorized into high (greater than median) and low (less than/equal to median) intake, dietary ITC was slightly lower in cases than controls but the difference was not significant [odds ratio (OR) 0.81, 95% confidence interval (CI) 0.59-1.12]. There were no overall associations between GSTM1, T1 or P1 genotypes and colorectal cancer risk. However, among individuals with both GSTM1 and T1 null genotypes, we observed a 57% reduction in risk among high versus low consumers of ITC (OR 0.43, 95% CI 0.20-0.96), in particular for colon cancer (OR 0.31, 0.12-0.84). Our results are compatible with the hypothesis that ITCs from cruciferous vegetables modify risk of colorectal cancer in individuals with low GST activity. Further, this gene-diet interaction may be important in studies evaluating the effect of risk-enhancing compounds in the colorectum.