Considerable discrepancies in the previously reported associations of the C-reactive protein (CRP) gene variants and ischemic stroke (IS) risk prompted us to perform this meta-analysis. We selected the fixed effects Mantel-Haenszel method to estimate the risk of IS [OR (odds ratio) along with its 95 % CI (confidence interval)] in relation to the CRP variants (-717 A > G, 1444 C > T). Heterogeneity test, influence analysis and publication bias test were appropriately performed using respective methods. We analyzed 1,926 IS patients and 2,678 controls and found the -717 A > G variant was not significantly associated with overall IS risk. Subsequent analysis of the 1444 C > T variant involving 3,278 samples similarly revealed no significant association with IS. There was no substantial heterogeneity or publication bias in this analysis. Our meta-analysis may provide first evidence showing that genetic variants within the CRP locus are unlikely to modulate risk of IS.