Aims/hypothesis In this study, we aimed to examine the association between age at natural menopause and risk of type 2 diabetes, and to assess whether this association is independent of potential mediators. Methods We included 3639 postmenopausal women from the prospective, population-based Rotterdam Study. Age at natural menopause was self-reported retrospectively and was treated as a continuous variable and in categories (premature, <40 years; early, 40–44 years; normal, 45–55 years; and late menopause, >55 years reference). Type 2 diabetes events were diagnosed on the basis of medical records and glucose measurements from Rotterdam Study visits. HRs and 95% CIs were calculated using Cox proportional hazards models, adjusted for confounding factors; in another model, they were additionally adjusted for potential mediators, including obesity, C-reactive protein, glucose and insulin, as well as for levels of total oestradiol and androgens. Results During a median follow-up of 9.2 years, we identified 348 individuals with incident type 2 diabetes. After adjustment for confounders, HRs for type 2 diabetes were 3.7 (95% CI 1.8, 7.5), 2.4 (95% CI 1.3, 4.3) and 1.60 (95% CI 1.0, 2.8) for women with premature, early and normal menopause, respectively, relative to those with late menopause ( p trend  <0.001). The HR for type 2 diabetes per 1 year older at menopause was 0.96 (95% CI 0.94, 0.98). Further adjustment for BMI, glycaemic traits, metabolic risk factors, C-reactive protein, endogenous sex hormone levels or shared genetic factors did not affect this association. Conclusions/interpretation Early onset of natural menopause is an independent marker for type 2 diabetes in postmenopausal women.