Significant improvements were made during the last two decades in the treatment of IDDM patients. To assess the risk of ESRD in a population that was exposed to these improvements, we determined the cumulative incidence of ESRD in a cohort of 142 white patients who were aged less than 21 years when they came to the Joslin Diabetes Center in 1959 with recently diagnosed IDDM. The first case of ESRD occurred after 13 years of IDDM, and a total of 25 cases have developed by 35 years' duration (cumulative incidence 21.3%). Median survival after the diagnosis of ESRD for the 16 patients who began dialysis was only 3.5 years. A strong predictor of the development of ESRD was the level of glycemic control during the first two decades of IDDM. ESRD developed in 36.3% of patients in the worst tertile for glycemic control but only in 14.4% and 9.2% of those in the middle and best tertiles. In comparison with two population based studies, the onset of ESRD in the Joslin Cohort was postponed by about five years. This advantage is more plausibly attributable to differences arising after the diagnosis of diabetes than to referral of less severe cases of IDDM to the Joslin Diabetes Center. What differences in diabetes care accounted for the postponement of ESRD cannot be discerned from comparisons among published studies, but likely candidates include Joslin's long-term advocacy of good glycemic control and the prompt implementation of new clinical interventions, such as antihypertensive treatment.