To examine whether exposure to oral contraceptives (OCs) is associated with bone mineral density (BMD) in young women, we studied, cross-sectionally and longitudinally, 216 white and 260 black women enrolled in the Coronary Artery Risk Development in Young Adults (CARDIA) study. Spine, hip and whole body BMDs were measured by dual-energy X-ray absorptiometry (DXA) when the women were aged 25-37 years, and whole body BMD was remeasured in 369 of the women 3 years later. A comprehensive history of OC use, including dose of ethinyl estradiol (estrogen) and duration of use, was determined from an interviewer-administered questionnaire. After adjustment for other relevant variables, we found that cumulative estrogen from OCs (mg) explained 4.0% of the variation in spine BMD (p = 0.024) among white women, but did not explain any of the variance in BMD in black women. Cumulative OC estrogen was associated with a decreased risk for low bone density (lowest quartile) at the spine, hip and whole body in white women. The odds ratios (95% Cls) comparing women in the highest quartile of cumulative OC estrogen with those in the lowest quartile were, at the spine: 0.08 (0.02, 0.46); at the hip: 0.23 (0.06, 0.87); and at the whole body: 0.37 (0.11, 1.26). OC exposure was not related to low bone density in black women. OCs did not predict longitudinal changes in whole body BMD in either race. These results suggest that exposure to the estrogen from OCs during the premenopausal years may have a small beneficial effect on the skeleton in white women. Benefit is proportional to the cumulative estrogen exposure, suggesting that previous cross-sectional studies that considered OC use as a dichotomous variable may have lacked the power to detect an association.