As part of systematic treatment trials of haloperidol, clozapine, and olanzapine with a total of 35 children and adolescents with early onset psychosis, prolactin was measured at baseline and week 6 of treatment. The National Institute of Mental Health patients--13 females, 22 males (mean age, 14.1+/-2.3 years; range, 9.1-19 years) with childhood onset schizophrenia (n = 32), or Psychotic Disorder not otherwise specified (NOS) (n = 3) with onset of psychosis before age 13--were recruited for open or double-blind trials of haloperidol, clozapine, or olanzapine. Baseline serum prolactin was measured after a 3-week washout period and after 6 weeks of treatment. Mean prolactin concentration after 6 weeks of treatment was significantly elevated on all three drugs; however, on clozapine, mean prolactin remained within the normal range. Prolactin was increased above the upper limit of normal for 100% of 10 patients on haloperidol, 70% of 10 patients on olanzapine, and 0% of 15 patients on clozapine (chi2 analyses: H > C, p = 0.004; O > C, p = 0.001). Given the potential endocrine and possible cardiac correlates of hyperprolactinemia, these more robust prolactin elevations in pediatric patients after short-term exposure to olanzapine than those reported for adults justify longer observation intervals with bigger samples to establish treatment safety of atypical antipsychotics in adolescents.