Estrogens exert their effect on the breast through the estrogen receptor. We prospectively investigated breast cancer risk associated with 2 polymorphic sites in the estrogen receptor alpha gene (ESR1). A total of 4,248 Caucasian women from the Study of Osteoporotic Fractures were genotyped for the −401 T/C and −354 A/G polymorphisms in ESR1. Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the associations between genotypes and breast cancer. During a mean follow-up of 12.4 years, 252 (5.9%) women developed breast cancer. The HR (95% CI) for breast cancer were 0.928 (0.708, 1.22) and 0.834 (0.538, 1.29) for the −354 A/G and A/A genotypes, respectively. Interactions with −354 variant were observed for smoking (HR = 1.52 and 1.56 for A/G and A/A smokers, respectively; HR = 0.74 and 0.60 for A/G and A/A non-smokers, respectively; interaction p = 0.03) and walking (HR = 0.75 and 1.15 for A/G and A/A walkers, respectively; HR = 0.18 and 0.49 for A/G and A/A non-walkers, respectively; interaction p = 0.01). There were no differences in the HR for the −401 T/C genotypes. An interaction between parity and carriage of the T allele was found (HR = 0.60 vs. 1.12 for nulliparous vs. parous women; interaction p = 0.03). ESR1 polymorphisms in combination with lifestyle factors may be associated with breast cancer risk in older Caucasian women. © 2005 Wiley-Liss, Inc.