Susceptibility to gastric cancer may be in part attributable to inter-individual variability in metabolic activation or detoxification of carcinogens, and in this context the polymorphic GSTM1 gene has been extensively studied. Seventeen reports detailing a possible association between GSTM1 deletion and gastric cancer have been published so far. In order to examine the risk of gastric cancer associated with GSTM1 null genotype, a meta-analysis of published case-control studies was undertaken using a random effect model. Two studies were excluded because some data were missing in the results. The principal outcome measure was the odds ratio for the risk of gastric cancer. Pooling all the 15 studies identified, the overall odds ratio of gastric cancer risk associated with GSTM1 deficiency was 1.24 (95% confidence interval, 1.00-1.54). Furthermore, by pooling four studies detailing the possible interaction between GSTM1 status smoking habits and gastric cancer risk, an overall estimate of odds ratio of 2.93 (95% confidence interval, 1.56-5.47) for ever smokers with GSTM1 deficiency compared to non-smokers with GSTM1 normal genotype has emerged. These results suggest that GSTM1 status probably has no effect on the risk of gastric cancer per se, but may modulate tobacco-related carcinogenesis of gastric cancer. Greater attention should therefore be paid to the design of future studies: only well designed population-based control studies considering all the possible confounding risk factors and based on a sample size commensurate with the detection of small genotypic risk may allow a more definitive conclusion.