Attempts to mobilize a more aggressive national prevention and treatment program for diabetes mellitus have been slow to develop and have been compromised by the absence of population-based data. This study used population-based Manitoba Health Claims data to measure diabetes prevalence, hospital utilization and mortality for the fiscal year 1991-92. A cohort of 45,117 persons with diabetes were identified, of whom 20,496 (45%) were hospitalized. Standardized comparisons between the diabetic and non-diabetic population revealed a greater than two fold increase in the use of hospital beds by persons with diabetes. This increased bed use was related to higher rates of admission, readmission and longer lengths of stay. Excess diabetes hospitalization was pervasive and hospital-wide in its scope, and remained consistent regardless of length of stay, discretionary nature of admitting diagnosis, intensity or complexity of care, region or treaty status. Analysis of complication-specific hospital bed use demonstrated that admissions for cardiovascular disease, stroke and extremity complications accounted for much of the excess hospitalization in the diabetic population, particularly in rural and Aboriginal populations. Assessment of mortality demonstrated that diabetes was associated with a two-fold excess premature mortality. This methodology provides a rapid, cost-effective means by which diabetes burden may be identified and subsequently targeted.