In 1993-1995, the authors evaluated risk factors for elevated blood and bone lead levels in 264 Boston, Massachusetts, area women previously selected for a case-control study of lead and hypertension. Bone lead was measured at the tibia and patella with K x-ray fluorescence. Blood lead was analyzed by graphite furnace atomic absorption. Participants were aged 46-74 years and had mean lead levels of 3 (standard deviation, 2) micro g/dl (blood), 13 (standard deviation, 9) micro g/g (tibia), and 17 (standard deviation, 11) micro g/g (patella). In multivariate linear regression models, use of postmenopausal estrogen (inverse) and alcohol intake (positive) were significantly associated with blood lead levels. Both bone lead measures were significantly and positively associated with blood lead but only among postmenopausal women not using estrogen; for example, an increase from the first to the fifth quintile of tibia lead level (19 micro g/g) was associated with a 1.7- micro g/dl increase in blood lead (p = 0.0001) in this group. Older age and lower parity were associated with higher tibia lead; only age was associated with patella lead. The observed interaction of bone lead with estrogen status in determining blood lead supports the hypothesis that increased bone resorption, as occurs postmenopausally because of decreased estrogen production, results in heightened release of bone lead stores into blood.