This study examined how well private-practice pediatricians can identify emotional/behavioral problems among preschool children. Children aged 2 through 5 (N = 3876) were screened during a visit to 1 of 68 pediatricians who rendered an opinion about the presence of emotional/behavioral problems. Subsequently, children who scored above the 90th percentile for behavioral problems on the Child Behavior Checklist, along with children matched on age, sex, and race who had screened low, were invited for an intensive second-stage evaluation. There were 495 mothers and children who participated in that evaluation, which included a behavioral questionnaire, maternal interview, play observation, and developmental testing. Two PhD-level clinical child psychologists rendered independent opinions about the presence of an emotional/behavioral disorder. The psychologists identified significantly higher rates of problems overall--13.0% when the criterion was independent agreement that the child had an emotional/behavioral problem and a regular psychiatric diagnosis was assigned, vs 8.7% based on pediatricians' ratings. Prevalence rates based on psychologists' independent ratings were significantly higher than pediatricians' for both sexes, 4- through 5-year-olds, and whites, but not for 2- through 3-year-olds, African-Americans, and all minorities. Prevalence rates based on psychologists' ratings were significantly higher than the pediatricians' for all subgroups when V-code diagnoses were included in the psychologists' ratings. Overall, pediatricians' sensitivity was 20.5%, and specificity was 92.7%. At least 51.7% of the children who had an emotional/behavioral problem based on the psychologist's independent agreement had not received counseling, medication, or a mental health referral from the pediatrician.(ABSTRACT TRUNCATED AT 250 WORDS)