Objective: This study followed remitted patients from a randomized controlled trial of adults with major depressive disorder (MDD). The aims were to describe rates of recurrence and to evaluate 3 clinical predictor domains. Method: Ninety-four treatment-naïve patients (50% female; Mage = 38.1 years; 48.9% White; 30.9% Hispanic) with MDD who had remitted to 12-week monotherapy (escitalopram, duloxetine, or cognitive behavior therapy [CBT]) participated in a 21-month maintenance phase (i.e., continued medication or 3 possible CBT booster sessions per year). Recurrence was assessed quarterly, and the clinical predictors were the following: 2 measures of residual depressive symptoms, 1 measure of lifetime depressive episodes, and 2 measures of baseline anxiety. Survival analysis models evaluated recurrence rates, and regression models evaluated the predictors. Results: Among all patients, 15.5% experienced a recurrence, and the survival distributions did not statistically differ among treatments. Residual depressive symptoms on the Hamilton Depression Rating Scale at the end of monotherapy were associated with increased risk for recurrence (hazard ratio = 1.31, 95% confidence interval [CI: 1.02, 1.67], Wald χ2 = 4.41, p = .036), and not having a comorbid anxiety disorder diagnosis at study baseline reduced the risk of recurrence (hazard ratio = .31, 95% CI [.10, .94], Wald χ2 = 4.28, p = .039). Conclusions: The study supported the benefits of maintenance treatment for treatment-naïve patients who remitted to initial monotherapy; nevertheless, remitted patients with a comorbid anxiety disorder diagnosis at the beginning of treatment or residual depressive symptoms after initial treatment were at risk for poorer long-term outcomes.