Aims: To investigate the association of several single-nucleotide polymorphisms (SNPs) within methylenetetrahydrofolate reductase (MTHFR) gene, and additional gene–environment interaction with ischemic stroke (IS) risk. Methods: Testing for Hardy–Weinberg equilibrium in controls was conducted using SNPstats (online software: http://bioinfo.iconcologia.net/SNPstats ). Generalized multifactor dimensionality reduction (GMDR) was used to screen the best interaction combination among four SNPs within MTHFR gene and smoking or alcohol drinking. Results: The frequency of the rs4846049-A allele was 28.6% in IS patients and 19.1% in normal controls, in addition, the frequency of the rs3737967-T allele was 27.9% in IS patients and 20.3% in normal controls, which was also indicating a statistically significant difference. The rs4846049-A and rs3737967-T were associated with an increased risk of IS risk; adjusted odds ratios (ORs) (95% confidence interval [CI]) were 1.76 (1.28–2.13) and 1.51 (1.13–1.97), respectively. GMDR model found significant gene–alcohol drinking interaction combination, but no significant gene–tobacco smoking interaction combinations. In order to obtain the odds ratios and 95% CI for the joint effects of gene–alcohol drinking on IS, we conducted stratified analysis for interaction effect using logistic regression. We found that alcohol drinkers with rs4846049-CA/AA genotype also have the highest IS risk, compared with never drinkers with rs4846049-CC genotype, OR (95% CI) = 3.12 (1.83–4.45), after adjustment for age, smoke, and smoking status. Conclusions: The rs4846049-A and rs3737967-T, gene–environment interaction between rs1764391 and rs918592, gene–environment interaction between rs4846049 and alcohol drinking were all associated with increased IS risk.