Studies of long-term effects of breastfeeding on the health of both infants and mothers often rely on maternal recall of breastfeeding duration after several decades. The authors evaluated this recall by 140 college-educated, US women 69–79 years of age who breastfed a child in 1940–1956 and recorded the duration both prospectively in a diary for the Menstruation and Reproductive History Study and retrospectively in a questionnaire in 1990–1991. Mean prospective breastfeeding duration was 5.6 months (range, 1–12 months). Mean reporting difference, questionnaire minus diary duration, was 0.0 months, with a standard deviation of 2.7 months; women who recorded short durations tended to overreport, while women who noted long durations underreported. The weighted kappa statistic for reporting agreement was 0.55 (95% confidence interval: 0.42, 0.67), with better recall observed for women in the youngest quintile at recall, firstborns, and infants with more siblings. Ever having breastfed was recalled by 94% of women. For categories of 1–3, 4–6, 7–9, and 10–12 months, recalled breastfeeding duration was correctly classified by 54% of women and was classified within ±1 category by 89%. The observed misclassification, if nondifferential with respect to outcome, would appreciably attenuate estimates of dose-response associations between breastfeeding duration and later health.