Aims. To determine the risk, hazard rate and factors affecting progression to diabetes in monozygotic twins of patients with Type I (insulin-dependent) diabetes mellitus. Methods. Prospective analysis was done of two cohorts of non-diabetic monozygotic twins of patients with Type I diabetes from Great Britain (n = 134) and the United States (n = 53). Results. The diabetes-free survival analysis was similar between both cohorts (p = 0.6). The combined survival analysis (n = 187, median follow-up = 17.7 years, range = 0.01–57) at 40 years of discordance estimated a 39 % probability of diabetes for the initially discordant twin. Survival analysis with left truncation of data estimated that probability to be 50 %. For twins who became concordant (n = 47), the median discordance time was 4.2 years (range 0.4 to 39), exceeding 15 years in 23.4 %. Twins of probands diagnosed at 24 years of age or younger had a 38 % probability of diabetes by 30 years of discordance, compared with 6 % for twins of probands diagnosed after 24 years of age (p = 0.004). The twins of probands diagnosed before 15 years of age had the highest diabetes hazard rate in the first discordance year, decreasing thereafter. By survival analysis, diabetes risk was higher in twins who were heterozygous for DR3-DQ2 and DR4-DQ8 than in twins with neither DR3-DQ2 nor DR4-DQ8 (p < 0.05). Conclusion/interpretation. Monozygotic twins of patients with Type I diabetes from two different countries had similar rates of progression to diabetes. Whereas most twins did not develop diabetes, 25 % of the twins who progressed did so after more than 14 years of discordance. An age-related heterogeneity was observed, with higher progression to diabetes for twins of patients diagnosed at a younger age. [Diabetologia (2001) 44: 354–362]