BACKGROUND Several studies have shown an increase of mortality in diabetic patients treated with combinations of sulphonylureas and biguanides. Comparisons between different insulin secretagogues in combination with metformin with respect to all-cause mortality have not been reported so far. METHODS An observational cohort study was performed on a consecutive series of 2002 outpatients with type 2 diabetes mellitus. Of these patients, 696 (34.8%) were receiving combinations of insulin secretagogues and biguanides at enrollment. Three-year mortality was assessed through research in the City of Florence Registry Office. RESULTS During follow-up, 295 deaths were recorded. Among patients on combined secretagogue and biguanide treatment, glibenclamide was associated with a significantly higher yearly mortality (8.7%) than repaglinide (3.1%; p = 0.002), gliclazide (2.1%; p = 0.001), and glimepiride (0.4%; p < 0.0001). After adjusting for potential confounders (including age; duration of diabetes; Body Mass Index (BMI); lipid profile; HbA(1c); insulin treatment; metformin doses; Charlson co-morbidity score; CCS), mortality remained significantly higher in patients treated with combinations of glibenclamide and metformin when compared to those treated with different insulin secretagogues (OR with 95% CI: 2.09 [1.07;4.11]). CONCLUSIONS In the present study, sulphonylureas with greater selectivity for beta-cell receptors, such as glimepiride and gliclazide, were associated with a lower mortality when used in combination with metformin in comparison with glibenclamide. Safety of such combinations deserves further investigation.