Stress is hypothesized to be a risk factor for breast cancer. The authors examined associations of hours of, and self-reported levels of stress from, informal caregiving with prospective breast cancer incidence. Cross-sectional analyses of caregiving and endogenous sex steroid hormones were also conducted. In 1992 or 1996, 69,886 US women from the Nurses' Health Study, aged 46-71 years at baseline, answered questions on informal caregiving; 1,700 incident breast cancer cases accrued over follow-up to 2000. A subset of 665 postmenopausal women not taking exogenous hormones returned a blood sample in 1990. Numbers of hours of care provided to an ill adult or to a child were each summed and analyzed as 0 (reference), 1-14, and >/=15 per week. Cox proportional hazards models were used in prospective analyses and linear models in cross-sectional analyses. High numbers of caregiving hours and self-reported stress did not predict a higher incidence of breast cancer. However, compared with women providing no adult care, women providing >/=15 hours of adult care (median, 54) had significantly lower levels of estradiol (geometric mean, 9.21 pg/ml vs. 7.46 pg/ml (95% confidence interval: 6.36, 8.76)) and bioavailable estradiol (geometric mean, 1.86 pg/ml vs. 1.35 pg/ml (95% confidence interval: 1.00, 1.82)). Stress from caregiving did not appear to increase breast cancer risk.