Combined oral contraceptives (COC) use is a unique risk factor for stroke in women, and may modify the associations between genetic polymorphisms and stroke. To investigate whether the genetic variants identified in a recent genome-wide association study (GWAS) could be replicated in Chinese women, as well as, whether related risk was different in COC users, 451 stroke cases and 831 age- and region-matched controls were recruited from our cohort. Genotyping of 3 SNPs (rs700651, rs10958409, and rs1333040) was performed by the polymerase chain reaction assay with TaqMan probes. The history of contraceptive use and relevant information were obtained from a face-to-face interview. Odds ratios (OR) with 95 % confidence interval (CI) were estimated under conditional logistic regression model after adjustment for cardiovascular covariates. Our study replicated the associations of rs10958409 and rs1333040, with the risk of stroke, especially hemorrhagic subtype, but failed to confirm association of rs700651. COC use was associated with a 1.56-fold (OR 1.56, 95 % CI 1.21-2.01) increased risk of stroke. COC users with rs10958409 GA/AA or rs1333040 CT/TT genotypes had an increased risk of overall stroke by 1.59-fold (OR 2.59, 95 % CI 1.59-4.19) and 3.24-fold (OR 4.24, 95 % CI 1.71-10.49), respectively, compared with the non-users with wild-type genotypes. Moreover, the risk of hemorrhagic stroke increased by 4.81- and 15.06-fold when risk allele carriers of rs10958409 or rs1333040 who took COC. Our results confirmed the associations of two GWAS SNPs, also suggested combination effects of these genetic variants and COC use on stroke risk.