Available data about mortality of type 2 diabetic patients treated with different sulphonylureas are scarce and contradictory. We evaluated the associations between all-cause and cause-specific mortality and treatments with different sulphonylureas in a retrospective cohort of type 2 diabetic patients from a diabetes clinic. All 1277 patients treated with sulphonylureas during 19961997 were enrolled: 159 patients were treated with tolbutamide, 977 glibenclamide and 141 gliclazide. The baseline data (centralised laboratory parameters, anthropometric data and presence of chronic complications) were abstracted from the clinical records. Information on vital status was collected from demographic files after 14-year follow-up. Adjusted hazard ratios (HR) were estimated with Cox (all-cause mortality) or Fine and Gray models (cause-specific mortality), including several potential confounders. Five hundred and fifty-six patients died during the follow-up: 262 from cardiovascular causes, 158 from cancer and 136 from other causes. When compared with the glibenclamide users, the gliclazide and tolbutamide users showed a significantly lower cancer mortality (HR=0.30; 95% CI 0.16-0.55, and HR=0.48; 95% CI 0.29-0.79 respectively). These results were strongly confirmed in the 555 patients on sulphonylurea monotherapy. None of the patients who were treated with gliclazide monotherapy died from cancer during the follow-up, and the patients on tolbutamide treatment exhibited a lower cancer mortality than the glibenclamide users (HR=0.40; 95% CI 0.22-0.71). Data did not change after stratification for the duration of sulphonylurea treatment from diabetes diagnosis to the study enrollment. Cancer mortality was markedly reduced in the patients on gliclazide and tolbutamide treatment. These results suggest additional benefits for these drugs beyond their blood glucose-lowering effect and strongly advocate for further investigation.