OBJECTIVES: To investigate the effect of insulin glargine (glargine) versus NPH insulin (NPH) on long-term outcomes in Type 2 diabetes using the Diabetes Mellitus Model (DMM). METHODS: The DMM predicts numerous short- and long-term complications over 10 years, based on published studies. The main influence on outcomes is HbA1c, which is simulated over time. The version of the DMM used in this analysis takes into account results of a regression analysis (pooled glargine versus NPH clinical trials), which demonstrates a more favourable relationship between HbA1c decrease and hypoglycaemia incidence for glargine versus NPH. Different cohorts of 10,000 patients with Type 2 diabetes taking glargine or NPH were assumed. The three scenarios tested differed in target HbA1c attainable by glargine versus NPH (scenario 1: ? = 0.13%; scenario 2: ? = 0.44%; scenario 3: ? = 0.85%), corresponding to realistic possible improvements with comparable hypoglycaemia. Assumptions were based on clinical trials for scenarios 1 and 2, and on the regression analysis for scenario 3. RESULTS: The following relative risks (RR; glargine/NPH) were obtained for scenarios 1, 2 and 3 respectively: 0.97, 0.89 and 0.81 for long-term microvascular complications (need for renal dialysis: 0.97, 0.88 and 0.79); 0.99, 0.95 and 0.91 for long-term macrovascular complications (non-fatal myocardial infarction: 0.98, 0.95 and 0.88); 0.96, 0.88 and 0,79 for diabetic foot syndrome; and 0.99, 0.94 and 0.90 for mortality. The RR for a nocturnal hypoglycaemic event (0.81, 0.92 and 1.01) was consistent with the regression analysis. The RR reductions ranged from 1% in the less optimistic scenario to >20% in the 'best case' scenario. Varying mean baseline HbA1c and duration of diabetes in sensitivity analyses did not alter these outcomes. CONCLUSIONS: Assuming comparable hypoglycaemia, the better glycaemic control (HbA1c reduction) achieved with glargine than with NPH would lead to reduced long-term complications and mortality rate, based on this model.