IMPORTANCE Controlled studies have shown short-term efficacy of esketamine for treatment-resistant depression (TRD), but long-term effects remain to be established. OBJECTIVE To assess the efficacy of esketamine nasal spray plus an oral antidepressant compared with an oral antidepressant plus placebo nasal spray in delaying relapse of depressive symptoms in patients with TRD in stable remission after an induction and optimization course of esketamine nasal spray plus an oral antidepressant. DESIGN, SETTING, AND PARTICIPANTS In this phase 3, multicenter, double-blind, randomized withdrawal study conducted from October 6, 2015, to February 15, 2018, at outpatient referral centers, 705 adults with prospectively confirmed TRD were enrolled; 455 entered the optimization phase and were treated with esketamine nasal spray (56 or 84 mg) plus an oral antidepressant. After 16 weeks of esketamine treatment, 297 who achieved stable remission or stable response entered the randomized withdrawal phase. INTERVENTIONS Patients who achieved stable remission and those who achieved stable response (without remission) were randomized 1:1 to continue esketamine nasal spray or discontinue esketamine treatment and switch to placebo nasal spray, with oral antidepressant treatment continued in each group. MAIN OUTCOMES AND MEASURES Time to relapse was examined in patients who achieved stable remission, as assessed using a weighted combination log-rank test. RESULTS Among the 297 adults (mean age [SD], 46.3 [11.13] years; 197 [66.3%] female) who entered the randomized maintenance phase, 176 achieved stable remission; 24 (26.7%) in the esketamine and antidepressant group and 39 (45.3%) in the antidepressant and placebo group experienced relapse (log-rank P=.003, number needed to treat [NNT], 6). Among the 121 who achieved stable response, 16 (25.8%) in the esketamine and antidepressant group and 34 (57.6%) in the antidepressant and placebo group experienced relapse (log-rank P<.001, NNT, 4). Esketamine and antidepressant treatment decreased the risk of relapse by 51% (hazard ratio [HR], 0.49; 95% CI, 0.29-0.84) among patients who achieved stable remission and 70% (HR, 0.30; 95% CI, 0.16-0.55) among those who achieved stable response compared with antidepressant and placebo treatment. The most common adverse events reported for esketamine-treated patients after randomization were transient dysgeusia, vertigo, dissociation, somnolence, and dizziness (incidence, 20.4%-27.0%), each reported in fewer patients (<7%) treated with an antidepressant and placebo. CONCLUSIONS AND RELEVANCE For patients with TRD who experienced remission or response after esketamine treatment, continuation of esketamine nasal spray in addition to oral antidepressant treatment resulted in clinically meaningful superiority in delaying relapse compared with antidepressant plus placebo.