Genome-wide association studies (GWASs) have identified genetic variants associated with breast cancer. Most GWASs to date have been conducted in women of European descent, however, and the contribution of these variants as predictors in Japanese women is unknown. Here, we analyzed 23 genetic variants identified in previous GWASs and conducted a case-control study with 697 case subjects and 1,394 age- and menopausal status-matched controls. We fit conditional regression models with genetic variants and conventional risk factors. In addition, we created a polygenetic risk score, using those variants with a statistically significant association with breast cancer risk, and also evaluated the contribution of these genetic predictors using the c statistic. Eleven single-nucleotide polymorphisms (SNPs) revealed significant associations with breast cancer risk. A dose-dependent association was observed between the risk of breast cancer and the genetic risk score, which was an aggregate measure of alleles in seven selected variants, namely FGFR2-rs2981579, TOX3/TNRC9-rs3803662, C6orf97-rs2046210, 8q24-rs13281615, SLC4A7-rs4973768, LSP1-rs38137198, and CASP8-rs10931936. Compared to women with scores of 3 or less, odds ratios (ORs) for women with scores of 4-5, 6-7, 8-9, and 10 or more were 1.33 (95% confidence interval, 1.00-1.80), 1.71 (1.26-2.30), 3.01 (1.97-4.58), and 8.69 (2.75-27.5), respectively (P (trend) = 1.9 × 10(-9)). The c statistic for a model including the genetic risk score in addition to the conventional risk factors was 0.6933, versus 0.6652 with the conventional risk factors only (P = 1.3 × 10(-4)). Population-attributable fraction of the risk score was 33.0%. In conclusion, we identified a genetic risk predictor of breast cancer in a Japanese population. Risk models which include a genetic risk score are possibly useful in distinguishing women at high risk of breast cancer from those at low risk, particularly in the context of targeted prevention.