The aim of this study was to assess dose-related steady-state serum concentrations of olanzapine (OLZ) and its metabolites N-desmethyl OLZ (DMO) and 2-hydroxymethyl OLZ (2-OH-OLZ) (assessed by high-performance liquid chromatography) in 122 child and adolescent psychiatric patients (age 16.9 +/- 2.2, range, 10-21 years; 74 males, 48 females) with a variety of diagnoses: schizophrenia group (n = 80); nonschizophrenia group (n = 29); anorexia nervosa (AN) group (n = 13). Median OLZ serum concentrations were 32.7 (range, 1-118; all patients), 37.7 (2-115; schizophrenia group), and 18.7 (1-63, AN group) ng/mL. The median OLZ concentration-to-dose (C/D) ratio (n = 122) was 2.6, with 90% of the distribution between 0.8 and 5.5 (ng/mL)/(mg/d). OLZ concentration was significantly correlated with DMO (r = 0.567; P < 0.0005) but not with 2-OH-OLZ (r = 0.122; P = 0.188). Daily OLZ dose was correlated with OLZ concentration in all (r = 0.684; P < 0.0005), schizophrenic (r = 0.542; P < 0.0005), and AN (r = 0.805; P = 0.001) patients, respectively. Patients aged less than 16 years displayed similar C/D for OLZ (P = 0.58) but higher C/D for DMO (P = 0.003) than those 16 years or older. AN patients received lower median OLZ doses (7.5; 5-15 mg) than schizophrenic patients (12.5; 2.5-40 mg), even after correcting for body mass index (P = 0.02). OLZ dose did not differ (P = 0.088) between smokers and nonsmokers, but smokers showed lower C/D for OLZ than nonsmokers (P = 0.008). C/D for OLZ was 38% higher (P = 0.041) under comedication with selective serotonin reuptake inhibitors when compared with OLZ monotherapy. Multiple linear regression analysis revealed that 46% of the variation of OLZ concentration can be explained by dose, diagnosis, age, sex, smoking, and comedication. The data are compared with the literature, and the relevance of therapeutic antipsychotic drug monitoring in previously sparsely investigated subgroups, such as children and adolescents or patients with AN, is emphasized.