Objective: To test the validity and reliability of the Chinese Geriatric Mental State Schedule (CGMS) in Taiwanese elders. Methods: The CGMS has gone through a standardized two-way translation, a pretest phase, and consensus focus group meetings in order to modify relevant culture-related terms of the original English version. The interrater reliability of the CGMS among eight psychiatrists was conducted after a training course was given to them. Diagnoses generated by the CGMS-AGECAT (Automated Geriatric Examination for Computer Assisted Taxonomy) were compared with psychiatric diagnoses according to the DSM-III-R criteria. The sample subjects were aged 65 and over and recruited from a community (n = 36) and an “old age home” (n = 56). Results: Four of the eight diagnostic categories generated by the CGMS-AGECAT had a generalized kappa value of 1.0, and the figures for the remaining four categories were acceptable: .8 for depressive neurosis, .6 for anxiety disorder, .5 for schizophrenia, and .5 for depressive psychosis (generalized kappa = .5). The overall agreement between the CGMS-AGECAT and independent psychiatric diagnosis (based on the DSM-III-R criteria) was satisfactory. Conclusion: The CGMS has been found to be a crossculturally valid and reliable instrument for use in Taiwan.