Androgens have been hypothesized to influence risk of breast cancer through several possible mechanisms, including their conversion to estradiol and their binding to the estrogen receptor and/or androgen receptor (AR) in the breast. The CAG repeat polymorphism in AR exon 1 has been implicated in breast cancer risk; however, studies on the association between this polymorphism and breast cancer risk remain conflicting. In order to derive a more precise estimation of the relationship, a large population-based case-control study was performed. We found that a long CAG sequence has a protective effect on breast cancer using an a priori determined cutoff (< 22 or ≥ 22) in a dominant model analysis [SL-LL vs. SS, odds ratio (OR) = 0.86, 95% confidence intervals (CI): 0.67-1.10]. A similar result was obtained by analyzing seven detailed genotyping case-control studies by allele comparison in dominant and recessive models. However, larger scale primary study is required to further evaluate the interaction of AR CAG polymorphism and breast cancer risk.