This population-based retrospective study investigates temporal trends in adult-onset diabetes melirtus prevalence, incidence, and survival. The complete community-based medical records, including laboratory results, of all Rochester, Minnesota, residents with a clinical diagnosis of diabetes or diabetes-like condition were reviewed to identify incidence cases aged 30 years or more from 1945 to 1989 (n=1,847) and prevalence cases aged 45 years or more on January 1, 1970 (n=465), January 1, 1980 (n=689), or January 1, 1990 (n=973). Glucose values and case definitions were standardized throughout. Observed 10-year survival for 1970 and 1980 prevalence cases was compared with that expected for Minnesota white populations in 1970 and 1980, respectively. Age-adjusted prevalence rose 65% for men and 37% for women between 1970 and 1990. There were marked differences among prevalence groups in treatment type, the proportion diagnosed using glucose tolerance tests, and the proportion categonzed as obese. Relative survival for 1980 prevalence cases was not greater than that for 1970 prevalence cases. Age-adjusted incidence rates rose 47% for men and 26% for women between 1960 and 1965 and 1985 and 1989. These findings emphasize the need for heightened surveillance and intervention to reduce the burden of illness from adult-onset diabetes mellitus in the population. Am J Epidemiol 1997;146:12–22.