Epidemiological studies have evaluated the association between CYP17 MspA1 polymorphism and breast cancer risk. However, the results remain conflicting rather than conclusive. To derive a more precise estimation of the relationship, we performed this meta-analysis. Systematic searches of the PubMed and Medline databases were performed. A total of 35 studies including 22,090 cases and 28,498 controls were identified. Genotype distributions of CYP17 in the controls of all studies were in agreement with Hardy-Weinberg equilibrium (HWE) except for three studies. When all 35 studies were pooled into the meta-analysis, there was no evidence for significant association between CYP17 MspA1 polymorphism and breast cancer risk (for A1/A2 vs. A1/A1: OR = 1.00, 95% CI = 0.96-1.04; for A2/A2 vs. A1/A1: OR = 1.03, 95% CI = 0.97-1.08; for dominant model: OR = 1.01, 95% CI = 0.97-1.05; for recessive model: OR = 1.03, 95% CI = 0.98-1.08). In the subgroup analyses by ethnicity, menopausal status and source of controls, no significant associations were found in all genetic models. When sensitivity analyses were performed by excluding HWE-violating studies, all the results were not materially altered. In summary, the meta-analysis strongly suggests that CYP17 MspA1 polymorphism is not associated with increased breast cancer risk.