OBJECTIVE The present study determined interrater agreement on diagnoses achieved using the parent and child versions of the Anxiety Disorders Interview Schedule for Children for DSM-IV (ADIS-C/P) and examined informant, age, and gender influences on reliability. METHODS Diagnoses established for 153 seven- to 16-year-old children during live administration of the ADIS-C/P were compared to diagnoses identified by a second rater after viewing a video recording of the interviews. RESULTS When information from both parent and child interviews was used, the level of agreement between raters for principal diagnosis (kappa = .92) and the individual anxiety disorders (kappa = .80-1.0) was excellent. Agreement on common comorbid disorders was good (kappa = .65-.77). Agreement was also good to excellent when diagnoses were assigned based on separate child or parent interviews, aside from children's report of externalizing disorders. Age and gender did not consistently impact interrater agreement. CONCLUSIONS The data indicate that the present version of the ADIS-C/P provides consistent diagnostic results across different clinicians and indicates improvements in the reliability of diagnoses following criterion changes in DSM-IV.