At least fourteen cohort studies have documented an inverse association between coffee consumption and risk of type 2 diabetes. We examined the prospective association between coffee and tea consumption and the risk of type 2 diabetes mellitus among British men (n 4055) and women (n 1768) from the Whitehall II cohort. During 11·7 years follow-up there were a total of 387 incident cases of diabetes confirmed by self-report of doctor's diagnosis or glucose tolerance tests. Despite an inverse association between coffee intake and 2 h post-load glucose concentration at the baseline assessment, combined caffeinated and decaffeinated coffee (hazard ratio (HR) 0·80; 95 % CI 0·54, 1·18) or only decaffeinated coffee intake (HR 0·65; 95 % CI 0·36, 1·16) was not significantly associated with diabetes risk at follow-up after adjustment for possible confounders. There was an association between tea intake and diabetes (HR 0·66; 95 % CI 0·61, 1·22; P < 0·05) after adjustment for age, gender, ethnicity and social status, which was not robust to further adjustments. There was, however, an association between combined intake of tea and coffee (two or more cups per day of both beverage) and diabetes (HR 0·68; 95 % CI 0·46, 0·99; P < 0·05) after full adjustment. In conclusion, relatively moderate intake (more than three cups per day) of coffee and tea were not prospectively associated with incidence of type 2 diabetes although there was evidence of a combined effect. The limited range of exposure and beverage consumption according to socio-economic class may explain these conflicting findings.