This study investigated predictors and moderators of mood symptoms in the randomized controlled trial (RCT) of Multi-Family Psychoeducational Psychotherapy (MF-PEP) for childhood mood disorders. Based on predictors and moderators in RCTs of psychosocial interventions for adolescent mood disorders, we hypothesized that children's greater functional impairment would predict worse outcome, whereas children's stress/trauma history and parental expressed emotion and psychopathology would moderate outcome. Exploratory analyses examined other demographic, functioning, and diagnostic variables. Logistic regression and linear mixed effects modeling were used in this secondary analysis of the MF-PEP RCT of 165 children, ages 8 to 12, with mood disorders, a majority of whom were male (73%) and White, non-Hispanic (90%). Treatment nonresponse was significantly associated with higher baseline levels of global functioning (i.e., less impairment; Cohen's d = 0.51) and lower levels of stress/trauma history (d = 0.56) in children and Cluster B personality disorder symptoms in parents (d = 0.49). Regarding moderators, children with moderately impaired functioning who received MF-PEP had significantly decreased mood symptoms (t = 2.10, d = 0.33) compared with waitlist control. MF-PEP had the strongest effect on severely impaired children (t = 3.03, d = 0.47). Comprehensive assessment of demographic, youth, parent, and familial variables should precede intervention. Treatment of mood disorders in high-functioning youth without stress/trauma histories and with parents with elevated Cluster B symptoms may require extra therapeutic effort, whereas severely impaired children may benefit most from MF-PEP.