Under-reporting of diabetes on death certificates contributes to the unreliable estimates of mortality as a result of diabetes. The influence of obesity on mortality in Type 2 diabetes is not well documented. We aimed to study mortality from diabetes and the influence of obesity on mortality in Type 2 diabetes in a large cohort selected from the General Practice Research Database (GPRD). Methods A cohort of 44 230 patients aged 35-89 years in 1992 with Type 2 diabetes was identified. A comparison group matched by year of birth and sex with no record of diabetes at any time was identified (219 797). Hazards ratios (HRs) for all-cause mortality during the period January 1992 to October 1999 were calculated using the Cox Proportional Hazards Model. The effects of body mass index (BMI), smoking and duration of diabetes on all-cause mortality amongst people with diabetes was assessed (n = 28 725). Results The HR for all-cause mortality in Type 2 diabetes compared with no diabetes was 1.93 (95% CI 1.89-1.97), in men 1.77 (1.72-1.83) and in women 2.13 (2.06-2.20). The HR decreased with increasing age. In the multivariate analysis in diabetes only, the HR for all-cause mortality amongst smokers was 1.50 (1.41-1.61). Using BMI 20-24 kg/m(2) as the reference range, for those with a BMI 35-54 kg/m(2) the HR was 1.43 (1.28-1.59) and for those with a BMI 15-19 kg/m(2) the HR was 1.38 (1.18-1.61).