We conducted a case-control study of 207 lung cancer patients and 283 controls to estimate the association of the MspI polymorphism in the cytochrome P450 CYP1A1 gene with lung cancer. The analysis of the CYP1A1 gene polymorphism was performed by RFLP analysis of PCR-amplified DNA. The association of the CYP1A1 polymorphism with lung cancer was assessed by logistic regression using the generalized additive modeling technique to adjust for race, education, smoking status, pack years, time since quitting smoking, asbestos exposures, and family cancer history. The frequencies of the MspI homozygote and heterozygote variant genotypes of CYP1A1 were 1% and 17%, respectively, in both lung cancer patients and controls. A significant association was found between the combined heterozygous and homozygous MspI variant of the CYP1A1 gene and lung cancer; the estimated odds ratio was 2.08 (95% confidence interval, 1.15-3.73). The association remained significant when we excluded the homozygous MspI variant individuals, the non-Caucasians, or the long-time tobacco quitters, and it was not modified by gender or cumulative cigarette consumption. In a specific histological cell type analysis, a positive association was found for each subtype of lung cancer, with no appreciable difference between cell types. In summary, our study demonstrated that the MspI variant CYP1A1 genotype is significantly associated with an increased risk of lung cancer among Caucasians with the odds ratio approximately, equivalent to 2 after controlling for important confounding factors. These results are similar to those obtained from reanalysis of published data on the combined CYP1A1 genotypes in a Japanese population (combined odds ratio, 1.95; 95% confidence interval, 1.17-3.28). Moreover, the elevated risk is noted in heterozygotes, i.e., individuals who carry only one copy of the variant gene.