The potential effectiveness of imipramine hydrochloride (up to 5 mg/kg/d) was investigated in 53 prepubertal children suffering from major depressive disorder. Two complementary strategies were used simultaneously: a five-week, double-blind, placebo-controlled design (N = 38), and a plasma level/clinical response study (N = 30). Fifteen of the 16 children randomly assigned to active drug in the first study also participated in the second. Subjects were assessed using the Schedule for Affective Disorders and Schizophrenia for School Age Children and diagnosed according to unmodified Research Diagnostic Criteria. Response rates in the double-blind study were similar in both groups (imipramine, 56%; placebo, 68%). In the plasma level study, total maintenance plasma level (imipramine plus desipramine) was found to positively and linearly predict clinical response of the depressive syndrome (P less than .003). No evidence of a curvilinear relationship was found. Depressive hallucinations during the episode negatively predicted clinical response (P less than .05). Weight-corrected imipramine dosage did not predict either clinical response or plasma level in the individual subject. No predictors of response were found in the placebo group. These results suggest that the mean imipramine dosage was too low, and that future double-blind, placebo-controlled studies of imipramine in prepubertal major depression should include plasma level titration to above 150 ng/mL and an initial placebo washout period.