Objective: To investigate differences in continuation rates between contraceptive and therapeutic use of the levonorgestrel-releasing intrauterine system 52 mg (LNG-IUS) and factors associated with early removal.Methods: Study design: Retrospective consecutive cohort design.Cohort: Women with the insertion of the LNG-IUS for contraceptive or therapeutic use from 1 January 2006 through 1 January 2009 at the Zuyderland Medical Centre, The Netherlands, with a follow-up of 5 years. The continuation period and reasons of early removal were noted. Univariable and multivariable analysis were performed.Results: Follow-up was possible in 2481 women, 1855 (74.8%) in the contraception group, and 626 (25.2%) in the therapy group. Multivariable Cox proportional hazards models showed, that therapeutic use was associated with an increased risk of early removal of the LNG-IUS (HR 1.23; 95% CI 1.08–1.41), as was having one child (HR 1.20; 95% CI 1.04–1.38), and a decreased risk with advancing age (HR 0.96; 95% CI 0.95–0.97). In both groups, an unacceptable bleeding pattern and adverse events were the main reasons of early removal, resulting in very low continuation rates over the years.Conclusion: Therapeutic use, having one child and a younger age are independent risk factors of early removal of the LNG-IUS, in contrast with previous LNG-IUS use which is associated with a lower risk. In both the contraception group and the therapy group, the main reasons for LNG-IUS discontinuation are continuation with a new LNG-IUS, and no more need for an LNG-IUS (for contraception or therapy). An unacceptable bleeding pattern or adverse events are associated with the lowest continuation rates in both groups.