Aims/hypothesis Previous literature documents controversial results for the impact of dehydroepiandrosterone (DHEA) in glucose metabolism. We aimed to assess the associations between serum levels of DHEA and its main derivatives DHEA sulphate (DHEAS) and androstenedione, as well as the ratio of DHEAS to DHEA, and risk of type 2 diabetes. Methods We used data on serum levels of DHEA, DHEAS and androstenedione from 5189 middle-aged and elderly men and women from the prospective population-based Rotterdam Study. Type 2 diabetes was defined as a fasting blood glucose ≥7.0 mmol/l or a non-fasting blood glucose ≥11.1 mmol/l. Results During a median follow-up of 10.9 years, 643 patients with incident type 2 diabetes were identified. After adjusting for age, sex, cohort, fasting status, fasting glucose and insulin, and BMI, both serum DHEA levels (per 1 unit natural logtransformed, HR 0.76, 95% CI 0.67, 0.87) and serum DHEAS levels (per 1 unit natural log-transformed, HR 0.82, 95% CI 0.73, 0.92) were inversely associated with risk of type 2 diabetes in the total population. Further adjustment for alcohol, smoking, physical activity, prevalent cardiovascular disease, serum total cholesterol, use of lipid-lowering medications, systolic BP, treatment for hypertension, C-reactive protein, oestradiol and testosterone did not substantially affect the association between DHEA and incident type 2 diabetes (per 1 unit natural log-transformed, HR 0.80, 95% CI 0.65, 0.99), but abolished the association between DHEAS and type 2 diabetes. Androstenedione was not associated with risk of type 2 diabetes, nor was DHEAS to DHEA ratio. Conclusions/interpretation DHEA serum levels might be an independent marker of type 2 diabetes.