Objective: This study compared the psychometric properties of the Children's Depression Rating Scale-Revised (CDRS-R) and the Montgomery-Asberg Depression Rating Scale (MADRS) in children with major depressive disorder. Method: Children (N = 96; ages 8 to 11 years inclusive) with nonpsychotic major depressive disorder were enrolled. Participants were part of a multisite, outpatient, randomized, placebo-controlled, 9-week trial of fluoxetine (10 mg/day for the first week and 20 mg/day thereafter). The CDRS-R and MADRS were completed based on clinician interviews with both parents and children. Classic test theory and item response theory analyses were conducted. Results: The MADRS and CDRS-R total scores were correlated at baseline (r = 0.51) and at study exit (r = 0.85). Cronbach's [alpha] was 0.86 (CDRS-R) and 0.82 (MADRS) at exit. The effect sizes for change from baseline to exit between the fluoxetine and placebo groups were 0.78 (CDRS-R) and 0.61 (MADRS). There was agreement between the CDRS-R and MADRS in the declaration of treatment response (50% improvement from baseline to exit) in 84.2% of children. Test information function favored the CDRS-R. Conclusions: The CDRS-R showed greater effect size for differentiating drug and placebo and better test information than the MADRS in this study of depressed children. (Contains 1 figure and 6 tables.)