One hundred and thirty two episodes of diabetic non ketotic hyperosmolar states were studied after a prospective schedule of treatment was designed. The admission data, the prognostic factors and their outcome were analyzed. Initial high osmolarity, urea and sodium plasma levels and low plasma pH were related to the admission level of consciousness (p < 0.01). High glucose, osmolarity, urea and sodium plasma levels at entry were related to the admission level of dehydration (p < 0.01). In multivariate regression analysis, osmolarity was the most influential variable in both the level of consciousness and the admission level of dehydration (p < 0.0001). Twenty two patients died (16.9%). Septic shock was the most frequent cause of death (31%) and mortality was higher in patients with cardiovascular disease (acute myocardial infarction or stroke) as the precipitating factor for diabetic hyperosmolar state (p < 0.002). Older age, low blood pressure, low sodium, pH and bicarbonate plasma levels, and high urea plasma levels were related to mortality (p < 0.01). In multiple regression analysis, urea was the most influential mortality risk factor (p < 0.0118). Non survivors received higher doses of insulin than survivors (p < 0.01). All these data suggest that it is not the hyperosmolarity itself, but the hemodynamical state of the patients, which is the most influential factor on the prognosis of a diabetic hyperosmolar state.