Results from prospective studies on premenopausal serum hormone levels in relation to breast cancer risk have been inconclusive, especially with regard to tumor subtypes. Using a case-control study nested within the prospective European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (801 breast cancer cases and 1,132 matched control subjects), we analyzed the relationships of prediagnostic serum estradiol, free estradiol, progesterone, testosterone, free testosterone and sex hormone-binding globulin (SHBG) levels with the risk of breast cancer by estrogen and progesterone receptor-positive and -negative breast tumors and by age at diagnoses. Higher prediagnostic serum levels of testosterone and free testosterone were associated with an increased overall risk of breast cancer [ORQ4-Q1 = 1.56 (95% CI 1.15-2.13), ptrend = 0.02 for testosterone and ORQ4-Q1 = 1.33 (95% CI 0.99-1.79), ptrend = 0.04 for free testosterone], but no significant risk association was observed for estradiol, free estradiol, progesterone and SHBG. Tests for heterogeneity between receptor-positive and -negative tumors were not significant. When analysis were stratified by age at tumor diagnosis, the odds ratios observed for estradiol were stronger and borderline significant for breast cancer diagnosed at age less than 50 [ORQ4-Q1 = 1.32 (95% CI 0.87-2.01), ptrend = 0.05] compared to breast cancer diagnosed at age 50 or above [ORQ4-Q1 = 0.94 (95% CI 0.60-1.47), ptrend = 0.34, phet = 0.04]. In conclusion, our data indicate that higher premenopausal circulating testosterone levels are associated with an increased risk of developing breast cancer, but do not show a significant association of estradiol or progesterone with breast cancer risk, overall, by menstrual cycle phase or by tumor receptor status, although a possible risk increase with higher estradiol levels for tumors diagnosed before age 50 was seen.