The structure and content of the Munich-Composite International Diagnostic Interview (M-CIDI) for the assessment of DSM-IV symptoms, syndromes, and diagnoses is described along with findings from a test-retest reliability study. A sample of 60 community respondents were interviewed twice independently by trained interviewers with an average time interval of 38 days between investigations. Test-retest reliability was good for almost all specific DSM-IV core symptom questions and disorders examined, with kappa values ranging from fair for two diagnoses--bulimia (kappa 0.55) and generalized anxiety disorder (kappa 0.45)--to excellent (kappa above 0.72) for all other anxiety disorders and alcohol use disorders. Test-retest reliability for age of onset and time-related questions was fairly consistently high (intra-class correlation values of 0.79 or above), with one notable exception: the assessment of disorders with onset before puberty. We concluded that the M-CIDI is acceptable for respondents, efficient in terms of time needed for and ease of administration, and reliable in terms of consistency of findings over time periods of at least 1 month.