Context: Various anatomic brain abnormalities have been reported for attention-deficit/hyperactivity disorder (ADHD), with varying methods, small samples, cross-sectional designs, and without accounting for stimulant drug exposure. Objective: To compare regional brain volumes at initial scan and their change over time in medicated and previously unmedicated male and female patients with ADHD and healthy controls. Design, Setting, and Participants: Case-control study conducted from 1991-2001 at the National Institute of Mental Health, Bethesda, Md, of 152 children and adolescents with ADHD (age range, 5-18 years) and 139 age- and sex-matched controls (age range, 4.5-19 years) recruited from the local community, who contributed 544 anatomic magnetic resonance images. Main Outcome Measures: Using completely automated methods, initial volumes and prospective age-related changes of total cerebrum, cerebellum, gray and white matter for the 4 major lobes, and caudate nucleus of the brain were compared in patients and controls. Results: On initial scan, patients with ADHD had significantly smaller brain volumes in all regions, even after adjustment for significant covariates. This global difference was reflected in smaller total cerebral volumes (-3.2%, adjusted F1.280=8.30, P=.004) and in significantly smaller cerebellar volumes (-3.5%, adjusted F1.280=12.29, P=.001). Compared with controls, previously unmedicated children with ADHD demonstrated significantly smaller total cerebral volumes (overall F2.288=6.65; all pairwise comparisons Bonferroni corrected, -5.8%; P=.002) and cerebellar volumes (-6.2%, F2.288=8.97, P<.001). Unmedicated children with ADHD also exhibited strikingly smaller total white matter volumes (F2.288= 11.65) compared with controls (-10.7%, P<.001) and with medicated children with ADHD (-8.9%, P<.001). Volumetric abnormalities persisted with age in total and regional cerebral measures (P=.002) and in the cerebellum (P=.003). Caudate nucleus volumes were initially abnormal for patients with ADHD (P=.05), but diagnostic differences disappeared as caudate volumes decreased for patients and controls during adolescence. Results were comparable for male and female patients on all measures. Frontal and temporal gray matter, caudate, and cerebellar volumes correlated significantly with parent- and clinician-rated severity measures within the ADHD sample (Pearson coefficients between -0.16 and -0.26; all P values were <.05). Conclusions: Developmental trajectories for all structures, except caudate, remain roughly parallel for patients and controls during childhood and adolescence, suggesting that genetic and/or early environmental influences on brain development in ADHD are fixed, nonprogressive, and unrelated to stimulant treatment.