In this study, we investigated associations between susceptibility genes and cerebral infarctions in a Chinese population, and whether gene–gene interactions increase the risk of cerebral infarctions. Overall, 292 patients with cerebral infarctions and 259 healthy control individuals were included. Eight variants in five candidate genes were examined for the risk of stroke, including the SG13S32 (rs9551963), SG13S42 (rs4769060), SG13S89 (rs4769874), and SG13S114 (rs10507391) variants of the 5-lipoxygenase activating protein (ALOX5AP) gene, the G860A (rs751141) variant of the soluble epoxide hydrolase (EPHX2) gene, the A1075C (rs1057910) variant of the CYP2C9*2 gene, the C430T (rs1799853) variant of the CYP2C9*3 gene, and the A6986G (rs776746) variant of the CYP3A5 gene. Gene–gene interactions were explored using generalized multifactor dimensionality reduction methods. There were no statistically significant differences in the frequencies of the genotypes of the eight candidate genes. The generalized multifactor dimensionality reduction analysis showed a significant gene–gene interaction between SG13S114 and A6986G, with scores of 10 for cross-validation consistency and 9 for the sign test (P=0.0107). These gene–gene interactions predicted a significantly higher risk of cerebral infarction (adjusted for age, hypertension, and diabetes mellitus; odds ratio=1.80495%, confidence interval: 1.180–2.759, P=0.006). A two-loci gene interaction confers a significantly higher risk for cerebral infarction. The combinational analysis used in this study may be helpful in the elucidation of genetic risk factors for common and complex diseases.