The modulation of benzo[a]pyrene diolepoxide (BPDE)-DNA adduct levels by polymorphisms in the CYP1A1, GSTM1 and GSTT1 genes was assessed in leukocytes of Caucasian males. Eighty-nine coke oven workers (35 smokers, 36 ex-smokers and 18 non-smokers) were recruited from job categories with different exposure levels to polycyclic aromatic hydrocarbons (PAH), together with 44 power plant workers (all smokers) not exposed to PAH. BPDE-DNA adducts were detected in 69 of 133 (52%) DNA samples with a 100-fold variation (range 0.2-44 adducts/10(8) nt) and a median of 1.6 adducts/10(8) nt. All samples with the GSTM1 active genotype (n = 59) and five out of 74 samples with GSTM1*0/*0 (7%) showed non-detectable adducts (<0.2 adducts/10(8) nt) and 69 of 74 subjects with GSTM1*0/*0 (93%) had detectable adducts (>0.2 adducts/10(8) nt). The difference in adduct level between the GSTM1*0/*0 and GSTM1 active genotypes was highly significant (P < 0.0001). No significant difference in adduct level between the GSTT1*0/*0 and GSTT1 active genotypes was seen. All heterozygotes (CYP1A1*1/*2) from subjects of GSTM1 active type did not have detectable adducts. Among the GSTM1-deficient individuals (n = 69), 42 with the CYP1A1*1/*1 genotype showed a lower adduct level (median 1.3, range 0.2-4.1 adducts/10(8) nt) compared with 26 individuals with heterozygous mutated CYP1A1*1/*2 genotypes (median 2.5, range 0.4-6.1 adducts/10(8) nt, P < 0.015). One individual with low PAH exposure and the rare combination CYP1A1*2A/*2A-GSTM1*0/*0 showed an extremely high level of 44 adducts/10(8) nt. Significant differences in detectable adduct levels were found between the CYP1A1*1/*1 and CYP1A1*1/*2 genotypes in the exposed group low + medium (P = 0.01) and for all adduct levels, detectable and non-detectable (set at a fixed value), in highly exposed individuals and in ex-smokers (P = 0.03), whereas no such differences were observed in the control group. Mutated CYP1A1*1/*2 increased the adduct level in non-smokers from the exposed group (1.4 versus 2.2 adducts/10(8) nt), but had no effect on the smokers from the exposed group (2.3 versus 2.8 adducts/10(8) nt). When all variables were dichotomized, statistical evaluation showed that CYP1A1 status (P = 0.015), PAH exposure (P = 0.003) and smoking (P = 0.006) had significant effects on adduct levels which increased in the order: CYP1A1*1/*1 < CYP1A1(*1/*2 or *2A/*2A); environmental exposure < occupational exposure; non-smokers < smokers, whereby adducts increased with cigarette dose and the duration of smoking. Higher levels of BPDE-DNA adducts in individuals with the combined CYP1A1(1/*2 or *2A/*2A)-GSTM1*0/*0 genotype suggest that these genotype combinations are at increased risk for contracting lung cancer when exposed to PAH.