Background. The equivalency of the Composite International Diagnostic Interview delivered by human interviewers (CIDI) and its computerized version (CIDI-Auto) was examined for anxiety and depressive disorders. Methods. Subjects were 40 patients at an Anxiety Disorders Clinic and 40 general medical practice attenders. The CIDI-Auto and CIDI were administered in counterbalanced order on the same day and measures of computer attitudes and the acceptability of the two interview formats were also taken. Results. The CIDI-Auto and the CIDI were found to be equally acceptable to subjects on the dimensions of comfort and preference, while the CIDI-Auto was rated as less embarrassing but too long in comparison with the CIDI. The agreement between the two formats was acceptable with kappa values for ICD-10 diagnoses being above 0.65 and for DSM-III-R diagnoses above 0.5 except for two diagnoses (generalized anxiety disorder and dysthymia). Discrepancies between the two formats were predicted by computer attitudes and not by computer experiences or the tendency to respond in a socially desirable fashion. Conclusions. It is concluded that the CIDI-Auto in its self-administered form is an acceptable substitute for the CIDI for suitable subjects.