Epidemiological studies have evaluated the association between ATM 5557G>A (p.D1853N) 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 PubMed and Medline databases were performed. A total of nine studies included 3155 cases and 2752 controls were identified. When all nine studies were pooled into the meta-analysis, there was no evidence for significant association between 5557G>A mutation and breast cancer risk(for G/A vs. G/G: OR = 1.05, 95% CI = 0.83-1.34; for A/A vs. G/G: OR = 0.77, 95% CI = 0.58-1.03; for dominant model: OR = 1.04, 95% CI = 0.82-1.31; for recessive model: OR = 0.87, 95% CI = 0.69-1.09). In the subgroup analyses by family history and ethnicity, significant associations were found among Amerindians (for G/A vs. G/G: OR = 2.19, 95% CI = 1.38-3.47; for dominant model: OR = 2.15, 95% CI = 1.37-3.38). In summary, the meta-analysis suggest that ATM 5557G>A polymorphism is associated with increased breast cancer risk among Amerindians. However, due to the small subjects included in analysis and the selection bias existed in some studies, the results for Amerindians should be interpreted with caution.