Cigarette smoking is a risk factor for colon adenoma. The glutathione S-transferase enzymes are involved in the detoxification of carcinogenic compounds including those found in tobacco smoke, and thus, may be important modifiers of individual risk of developing this disease. We examined the prevalence of GSTM1 and GSTT1 gene deletions, and two GSTP1 polymorphisms in 772 cases with advanced colorectal adenomas (>1 cm, villous elements or high-grade dysplasia) of the distal colon (descending or sigmoid colon or rectum) and 777 sigmoidoscopy negative controls enrolled in the screening arm of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Epidemiologic data on smoking was collected by self-administered questionnaire and DNA was extracted from whole blood or buffy coat. For GSTM1 and GSTT1, we used a newly developed TaqMan-based assay capable of discriminating heterozygous (+/-) individuals from those with two active alleles (+/+) and homozygous deletions (-/-). For GSTP1, the I105V and the A114V substitutions were identified using end point 5' nuclease assays (TaqMan). Adjusted odds ratios (OR) and 95% confidence intervals (95% CI) were determined using unconditional logistic regression, controlling for age, race, and gender. Advanced adenoma risk was increased in current/former smokers (OR, 1.4; 95% CI, 1.2-1.8). Risks were decreased in subjects with > or =1 inactive GSTM1 alleles (OR, 0.6; 95% CI, 0.4-0.9); and the association was independent of smoking status (P interaction = 0.59). Having > or =1 inactive GSTT1 allele was associated with increased risk among smokers (OR, 1.4; 95% CI, 1.1-1.9; P trend = 0.02) but not among never smokers (OR, 0.9; 95% CI, 0.6-1.3) and a significant interaction between smoking and genotype was observed (P interaction = 0.05). In summary, this is the first study to report associations between colorectal adenomas and GSTM1 wild-type and GSTT1 null allele among smokers. These findings only became apparent using a newly developed assay able to distinguish heterozygous from wild-type individuals. Our data provide evidence that phenotypic differences between these two groups exist.