Background: Triglyceride-glucose index (TyG) is proposed as a prognostic index for mortality and for co-morbidities in the general population, in T2DM and in patients with cardiovascular diseases, but data on the predictive role of TyG for mortality in obesity are lacking. Methods. In a cohort of 1,992 obese patients (212 men and 1,780 women, aged 42.2±11.95 years, followed for 14.3±4.44 years), clinical and laboratory data were subdivided into age quartiles. Data were analyzed for the whole cohort, for non-diabetic and T2DM patients separately. A univariate regression analysis and Cox proportional-hazard models were implemented to evaluate the risk of death associated with various independent variables. Results: In the whole cohort, in T2DM and in non-diabetic patients there was a progressive deterioration in clinical/metabolic variables and a progressive increase of all-cause mortality with increasing age quartiles. In all patients, age and male sex were associated with increased mortality; in non-diabetic patients, CHD was associated with increased all-cause mortality. Higher TyG or TyG quartiles or median TyG, BG or BG quartiles or median BG were associated with increased all-cause mortality in the whole cohort and in T2DM patients. Metabolic syndrome was also associated with all-cause mortality, albeit with some differences. Conclusion: TyG is a risk factor for mortality in the whole cohort and in obese T2DM, but not in obese non-diabetic patients at any adult age. BG and metabolic syndrome also predict all-cause mortality. These data need to be confirmed in larger studies and with different ethnic groups.