OBJECTIVE To examine bone density among adolescents receiving different forms of hormonal contraception along with that of control subjects. METHODS Baseline and 1-year measures of lumbar vertebral bone density were obtained in girls receiving depot medroxyprogesterone acetate (Depo-Provera) (n = 15), levonorgestrel (Norplant) (n = 7), or oral contraceptives (n = 9) and in girls receiving no hormonal treatment (n = 17). In a subsample of Depo-Provera users (n = 8), Norplant users (n = 3), and control subjects (n = 4), bone density measurements were repeated after 2 years. Bone density was measured by dual-energy x-ray absorptiometry. RESULTS Body mass indexes, level of pubertal development, substance use, and reproductive histories were not significantly different among the groups. More black girls were represented in the initial Depo-Provera group (p < 0.02), girls in the Norplant group exercised more hours per week (p < 0.02), and control subjects were older (p < 0.01) than those in the other groups. These variables did not significantly affect bone density results. After 1 year, bone density decreased 1.5% in Depo-Provera users, compared with increases of 2.5% in Norplant users, 1.5% in oral contraceptive users, and 2.9% control subjects (p < 0.02). After 2 years, bone density increased a total of 9.3% in Norplant users and 9.5% in control subjects but decreased a total of 3.1% in Depo-Provera users (p < 0.0001). CONCLUSIONS These data suggest that Depo-Provera may, at least temporarily, suppress the expected skeletal bone mineralization in adolescents, whereas Norplant and oral contraceptives are associated with the expected increase in bone density in this population.