As contraceptive use increases, studies on contraceptive discontinuation have become increasingly important. This paper uses contraceptive calendar data collected in the 1991 and 1994 Indonesia Demographic and Health Surveys. It aims to study the factors associated with contraceptive discontinuation by reason in Bali, an Indonesian province successfully implementing family planning program. The cumulative probability of discontinuation at the first 12 months of the initiation of use in 1986-1991 was 19.3%, increased to 24.9% in 1989-94. Side effects and health concerns are the most common reasons to discontinue. With a multilevel discrete-time competing risks hazard model, the study found that contraceptive method chosen and duration of use are strongly associated with contraceptive discontinuation. Socio-economic and demographic factors are also important in this matter. Unobserved heterogeinity at woman level significantly affects contraceptive discontinuation.