Background: Continuous intra-duodenal infusion of levodopa-carbidopa intestinal gel (LCIG) is a well-established therapy for patients with advanced Parkinson's disease (PD) suffering from motor complications despite optimized treatment with oral dopaminomimetics. However, time to discontinuation of treatment with LCIG varies considerably between patients, ranging from a few months to more than ten years. To improve the selection of candidates for LCIG, knowledge of prognostic factors is of paramount importance. Objective: To explore baseline predictors of time to discontinuation of LCIG. Methods: In this two-center retrospective cohort study, we reviewed the medical files of 98 PD patients treated with LCIG between April 2006 and December 2015 (53% male; mean age: 66.2 years; mean disease duration: 12.3 years). Baseline patient characteristics were used as covariates in Cox regression models. Results: During follow-up (mean observation time: 2.6 years; range: 0.1-9.3) eighteen patients discontinued treatment (18.4%), while seven patients died (7.1%). Median duration of treatment with LCIG, estimated with Kaplan-Meier analysis, was 7.8 years (95% CI: 6.7-9.0). Disease duration (in years) at baseline was a statistically significant predictor of time to discontinuation of LCIG (HR: 0.85; 95% CI: 0.75-0.96, p = 0.006). All other characteristics studied, e.g. age >70 years, did not show statistically significant associations with the total duration of treatment with LCIG. Conclusion: Our findings show a low overall rate of discontinuation of LCIG infusion, with a median duration of treatment of 7.8 years. Shorter disease duration at baseline appeared to be a predictor of earlier discontinuation of LCIG.