OBJECTIVE: To examine the relationships between HDL cholesterol subfractions and the incidence of type 2 diabetes and to evaluate potential sex differences in these relationships. RESEARCH DESIGN AND METHODS: Proportional hazards analyses were performed to examine the relationships between HDL subfractions and the development of type 2 diabetes in Pima Indian women and men. Results were controlled for age, BMI, systolic blood pressure, and 2-h glucose. RESULTS: Some 54 of 123 women and 25 of 50 men developed type 2 diabetes during a mean follow-up of 10 (2-19) years. For women, in separate models, high levels of total HDL, HDL2a, and HDL3 were negatively associated with incidence of type 2 diabetes; results were unchanged in models further controlled for fasting insulin level or alcohol consumption. For men, the results were inconsistent and associated with wide confidence intervals; high total HDL and HDL3 were positively associated with incidence of type 2 diabetes in models further controlled for fasting insulin level, but the risk estimates were attenuated in models further controlled for alcohol consumption. CONCLUSIONS: High levels of total HDL, HDL2a, and HDL3 were potential protective factors against type 2 diabetes in women after accounting for alcohol consumption and insulin resistance. High levels of total HDL and HDL3 were predictive of type 2 diabetes in men; the relationship in men appeared to be due to an association with alcohol consumption. The sex differences in the effects of HDL cholesterol may be related to the effects of sex hormones or lipoproteins.