OBJECTIVE To determine Depo-Provera continuation rates and reasons for its discontinuation among adolescents. METHODS Medical record reviews and telephone interviews with 159 adolescents who initiated Depo-Provera use between 1 December 1992 and 31 December 1995 at two clinics in New York City. Depo-Provera continuation was measured using lifetable analysis. RESULTS The mean age was 17.7 +/- 1.5 years, with a median of 1 pregnancy (range 0-11). Mean follow-up was 23.4 +/- 10.7 months. Depo-Provera continuation rates were 71% at 3 months, 48% at 6 months, and 27% at 12 months, and were not affected by age, race, pregnancy or contraceptive history, clinic, or foster care status. Forty-three subjects (37% of discontinuers) restarted Depo-Provera during the study period, with a mean time to restart of 8.4 months after the last Depo-Provera injection. Side effects were the main reported reason for Depo-Provera discontinuation, primarily menstrual irregularities (26%) and weight gain (18%). Seventy percent of those discontinuing Depo-Provera owing to irregular bleeding did so after only one injection. For 23%, the single reason for discontinuation was appointment noncompliance. Restart rates were lowest among those who reported irregular bleeding (15%), weight gain (9%), and hair loss (10%), and highest among those discontinuing owing to missed appointments (87%) (p < 0.05). Pregnancies occurred in 19% of Depo-Provera discontinuers. CONCLUSIONS Although Depo-Provera continuation rates among adolescents are low, over a third of discontinuers may restart the method. Aggressive management of side effects and assistance with appointment follow-up may improve long-term use. High pregnancy rates warrant close follow-up after Depo-Provera discontinuation.