A list of admissions to a psychiatric hospital over 10 years (1972-81) was compared by computer with all Iowa death certificates for the same period. Each patient was assigned 1 psychiatric diagnosis from a hierarchical list of 10 categories. Ratios of observed to expected deaths, adjusted for age, sex & follow-up time, reveal relative risk for premature deaths. Advantages & disadvantages of this approach are discussed, as is a similar ongoing study of mortality in affective disorders. (KM)