Differences in human cancer susceptibility have been attributed to polymorphisms of carcinogen metabolizing enzymes. Our efforts have focused on the systems responsible for metabolism of aromatic and heterocyclic amines found in cigarette smoke and in cooked foods. Cytochrome P4501A2 (CYP1A2), which catalyzes aromatic and heterocyclic amine N-oxidation, has been implicated as a risk factor in both urinary bladder and colorectal cancer. In the present study we used the results of caffeine phenotyping experiments to measure the effects of cigarette smoke and compounds present in meat cooked at high temperature on CYP1A2 activity. Subjects in the smoking cessation study had mean CYP1A2 activity of 17.8 (expressed as the urinary molar ratio of [17X + 17U]/137X) while smoking: however, this activity decreased to 10.9 three weeks after cessation of smoking. Subjects in the cooked meat feeding study had mean CYP1A2 activity of 9.01 after 1 week of consuming meat cooked at low temperature, but this value increased to 12.7 after 1 week of consuming meat cooked at high temperature. Because no association has been identified between differences in CYP1A2 activity and variations in the CYP1A2 structural gene, we sought to determine whether the activities of other carcinogen metabolizing enzymes are involved in the regulation of CYP1A2 activity. CYP1A2 activity was higher in individuals who express the GSTM1 null allele compared to those expressing the GSTM1*A,B allele, 10.2 vs. 8.5 for unexposed conditions and 15.0 vs. 12.3 for exposed conditions. CYP1A1 genotyping demonstrated that individuals possessing the Ile/Ile CYP1A1 genotype had greater mean CYP1A2 activity than those who had the heterozygous Ile/Val allelic variant of the CYP1A1 gene. However, upon exposure to cigarette smoke or high-temperature cooked meat, individuals possessing the heterozygous form of the CYP1A1 gene had significantly increased CYP1A2 activity (18.1) compared to those with the more common Ile/Ile CYP1A1 genotype (13.3). These results indicate that CYP1A2, CYP1A1, and GSTM1 gene-gene interactions could be important confounders in the interpretation of molecular epidemiology studies.