Objective: It is biologically plausible for physical activity to decrease breast cancer risk; however, epidemiologic studies have yielded inconsistent findings. We therefore examined physical activity and breast cancer risk in the Women's Health Study. Methods: We assessed physical activity among 39,322 apparently healthy women, aged ≥45 years, and prospectively followed them for an average of 48 months. Four hundred eleven women developed breast cancer, with 222 positive for both estrogen and progesterone receptors. Results: Among all women the multivariate relative risks of all breast cancer associated with p-trend = 0.11. However, among postmenopausal women there was a significant inverse trend for all breast cancer; the corresponding relative risks were 1.0 (referent), 0.97 (0.68–1.4), 0.78 (0.54–1.1), and 0.67 (0.44–1.0), respectively; p-trend = 0.03. Physical activity was unrelated to breast cancers positive for both estrogen and progesterone receptors either among all or postmenopausal women (p-trend = 0.50 and 0.26, respectively). When we assessed only vigorous recreational activity, requiring ≥6 METs or multiples of resting metabolic rate, we observed no significant associations with all or steroid hormone receptor positive breast cancer, either among all or postmenospausal women. Conclusions: These data suggest that physical activity during middle age and older is not uniformly associated with decreased breast cancer risk. Among postmenopausal women only, higher levels of physical activity may decrease the risk of breast cancer. This study, however, had limited statistical power to detect small effects.