To evaluate the hypothesis that lactation stimulates lead release from bone to blood, the authors analyzed breastfeeding patterns and bone lead concentrations as determinants of blood lead levels among 425 lactating women in Mexico City for 7 months after delivery (1994-1995). The authors measured in vivo patella and tibia lead concentrations at 1 month postpartum using K x-ray fluorescence. Maternal blood samples and questionnaire information were collected at delivery and at 1, 4, and 7 months postpartum. Blood lead was analyzed using graphite furnace atomic absorption spectroscopy. Mean blood lead level at delivery was 8.4 microg/dl (range: 1.8--23.4). Mean cortical and trabecular lead levels were 10.6 microg/g (range: nondetectable to 76.5) and 15.3 microg/g (range: nondetectable to 85.9), respectively, reflecting a population with elevated and diverse past and current lead exposure. The association of bone lead and breastfeeding with blood lead was estimated using generalized estimating equations. Breastfeeding practices and maternal bone lead were important predictors of blood lead level. After adjustment for bone lead and environmental exposure, women who exclusively breastfed their infants had blood lead levels that were increased by 1.4 microg/dl and women who practiced mixed feeding had levels increased by 1.0 microg/dl, in relation to those who had stopped lactation. These results support the hypothesis that lactation is directly related to the amount of lead released from bone.