ABSTRACT: Importance:The ability of pharmacotherapies to prevent relapse and maintain efficacy with long-term treatment in psychiatric conditions is important. Objective:To assess lisdexamfetamine dimesylate maintenance of efficacy in adults with moderate to severe binge-eating disorder. Design, Setting, and Participants:A multinational, phase 3, double-blind, placebo-controlled, randomized withdrawal study including 418 participants was conducted at 49 clinical research study sites from January 27, 2014, to April 8, 2015. Eligible adults met DSM-IV-R binge-eating disorder criteria and had moderate to severe binge eating disorder (?3 binge-eating days per week for 14 days before open-label baseline; Clinical Global Impressions-Severity [CGI-S] scores ?4 [moderate severity] at screening and open-label baseline). Following a 12-week, open-label phase (dose optimization, 4 weeks [lisdexamfetamine dimesylate, 50 or 70 mg]; dose maintenance, 8 weeks), lisdexamfetamine responders (?1 binge eating day per week for 4 consecutive weeks and CGI-S scores ?2 at week 12) were randomized to placebo or continued lisdexamfetamine during a 26-week, double-blind, randomized withdrawal phase. Interventions:Lisdexamfetamine administration. Main Outcomes and Measures:The primary outcome variable, time to relapse (?2 binge-eating days per week for 2 consecutive weeks and ?2-point CGI-S score increases from randomized withdrawal baseline), was analyzed using a log-rank test (primary analysis); the analysis was stratified for dichotomized 4-week cessation status. Safety assessments included treatment-emergent adverse events. Results:Of the 418 participants enrolled in the open-label phase of the study, 411 (358 [87.1%] women; mean [SD] age, 38.3 [10.4] years) were included in the safety analysis set. Of 275 randomized lisdexamfetamine responders (placebo, n?=?138; lisdexamfetamine, n?=?137), the observed proportions of participants meeting relapse criteria were 3.7% (5 of 136) for lisdexamfetamine and 32.1% (42 of 131) for placebo. Lisdexamfetamine demonstrated superiority over placebo on the log-rank test (?21, 40.37; P?