OBJECTIVE: This study examined the reliability of symptom reporting by community children of elementary school age and their parents on a version of the Diagnostic Interview Schedule for Children-Revised (DISC- R). METHOD: A sample of 109 children aged 6-11 years from an ongoing epidemiologic study were recruited for retest DISC-R interviews after completion of the study protocol. Retest interviews took place 7-18 days after the first interview and were conducted by interviewers who had no prior information about the subjects. Test-retest reliability for five common childhood psychiatric diagnoses was evaluated with the kappa statistic; the intraclass correlation coefficient was used to evaluate test-retest reliability of symptom scales. RESULTS: The reliability of the parents' reports on the DISC-R was good to excellent for attention deficit hyperactivity disorder and separation anxiety disorder; it was fair for overanxious disorder, oppositional defiant disorder, and conduct disorder. The children reported many fewer symptoms than the parents except for separation anxiety disorder; reliability was fair for separation anxiety disorder and poor for attention deficit hyperactivity disorder. The children were particularly unreliable in reporting about time factors, such as duration and onset of symptoms. When symptoms were considered without duration and onset, children's reports reached fair reliability for separation anxiety disorder, overanxious disorder, and attention deficit hyperactivity disorder but remained poor for oppositional defiant disorder. CONCLUSIONS: The results suggest that highly structured diagnostic interviews such as the DISC-R may not be appropriate for use with younger children of elementary school age in community-based studies.