Interindividual differences in lung cancer susceptibility may be mediated in part through polymorphic variability in the bioactivation of procarcinogens. GSTM1 status has been extensively studied in this context as a lung cancer risk factor, although published studies have produced conflicting results. To clarify the impact of GSTM1 status on lung cancer risk a meta-analysis of 23 case-control studies from the literature has been carried out using a random effects model. The principal outcome measure was the odds ratio for the risk of lung cancer. There was heterogeneity between the studies attributable to differences in the methods of assigning GSTM1 status. Pooling the studies that were based on phenotyping methods, the overall odds ratio of lung cancer risk associated with GSTM1 deficiency was 2.12 (95% confidence interval, 1.43-3.13). The risk of lung cancer risk associated with GSTM1 deficiency derived from the studies based on genotyping methods was, however, lower. The overall odds ratio was 1.13 (95% confidence interval, 1.04-1.25). These findings suggest that the estimates of lung cancer risk associated with GSTM1 deficiency in the early studies, based on phenotyping, were overinflated. Moreover, it is conceivable, given publication bias, that GSTM1 status has no effect on the risk of lung cancer per se. A major concern in case-control studies of polymorphisms and cancer risk is bias. A review of the 23 case-control studies indicates that greater attention should, therefore, be paid to the design of future studies.