OBJECTIVE To establish all-cause and cause-specific death rates, and risk factors for mortality in insulin-treated diabetic individuals living in the province of Canterbury, New Zealand. METHODS Insulin-treated diabetic subjects (n = 995) on the Canterbury Diabetes Registry were followed up over 15 years and vital status determined. Death rates were standardized and hazard regression was used to model the effects of demographic covariates on relative survival time. RESULTS There were 419 deaths in 11 226.3 person-years of follow-up with a standardized mortality ratio (SMR) of 2.0 (95% confidence interval (CI) 1.8-2.2). Relative mortality was greatest for the group aged 0-29 years (SMR 3.0 (95% CI 2.4-3.7)). After controlling for diabetes duration and gender, a 10-year increment in age of onset was associated with a 33% decrease in relative hazard (95% CI 29-36%), indicating that excess mortality due to diabetes declines with rising age of onset. After controlling for age of onset and gender, each 10-year increment in duration of diabetes is associated with a 26% decrease in relative hazard (95% CI 24-29%), indicating that with longer survival the mortality hazard approaches the general population hazard. Relative mortalities were increased for cardiovascular, renal and respiratory disease, but not malignancy. Relative mortality from acute metabolic complications was increased in the subgroup with age of onset of diabetes < 30 years and requiring insulin within 1 year of diagnosis. CONCLUSIONS Mortality rates are high for insulin-treated diabetic individuals relative to the general population.