ABSTRACT: The main purpose of this study was to perform a meta-analysis to assess the interaction between smoking and nine genes (GSTM1, GSTT1, GSTP1, CYP1A1, NAT2, SULT1A1, hOGG1, XRCC1 and p53) on colorectal cancer, gastric cancer, liver cancer and oesophageal cancer. Published articles from the PubMed, ISI and EMBASE databases were retrieved. A total of 67 case-control studies or nested case-control studies were identified for the analysis. The pooled jodds ratio (OR) with 95% confidence interval (CI) was calculated using the random effect model. The overall study showed that the GSTM1 polymorphism was associated with the risk of the four digestive cancers among Asian population (OR 1.284, 95% CI: 1.122-1.470, p: 0). Subgroup analyses by cancer site showed that GSTM1 null genotype increased the gastric cancer risk in total population (OR 1.335, 95% CI: 1.145-1.556, p: 0). However, the association of GSTM1 null genotype with the oesophageal cancer risk was found in smokers (OR 1.382, 95% CI: 1.009-1.894, p:0.044), but not in non-smokers (OR 1.250, 95% CI: 0.826-1.891, p:0.290). Moreover, smokers with the CYP1A1 IIe462Val polymorphism were at an increased cancer risk in Asian population (OR=1.585, 95% CI 1.029-2.442, p: 0.037). None of the other gene-smoking interactions was observed in the above cancers. This meta-analysis reveals two potential gene-smoking interactions, one is between smoking and GSTM1 on oesophageal cancer, and the other is between smoking and CYP1A1 IIe462Val on the four cancers in Asian population. Future studies need to be conducted to verify the conclusions.