OBJECTIVES: To perform a beta-verification of a novel diabetes modeling and analysis framework (DMAF) designed to accommodate growing demand for analysis on ever-shifting special subpopulations, new interventions, and updated care algorithms. A Monte Carlo microsimulation model assuming standard oral and subsequent insulin therapy generated mean outcomes as defined by recently published trials: 1) ACCORD-BPLI; 2) ACCORD-GLI; 3) ASPEN; and  4) ADVANCE. METHODS: Diabetes is increasing in prevalence, and its 20-year history of diabetes care has witnessed a shift from treating complications to prevention based on evidence from: The United Kingdom Prospective Diabetes Study, Diabetes Control and Complications Trials, and the Wisconsin Epidemiological Study of Diabetic Retinopathy.  All have confirmed that tight control of hemoglobin A1c reduced the incidence of complications. Recent trials of diabetics have evaluated targeted interventions for clinical factors and impact on complication rates. Evidence from these trials suggests that aggressive A1c targets may not be suitable for all patients. The evolution of decision models for diabetes has paralleled that of care. Increased prevalence has placed pressure on health care costs and expectations that new interventions impart significant benefits. New evidence has in turn motivated development of decision models that evaluate new interventions, treatments, and care algorithms. RESULTS: The DMAF was reasonably consistent with well-defined composite endpoints for ASPEN (15.0% vs. 17.1%); ADVANCE: Secondary (10.5% vs. 9.6%), fatal MI (5.7% vs. 5.5%), all coronary events (11.9% vs. 10.3%); and ACCORD-BPLI non-fatal MI (1.4% vs. 1.3%). DMAF showed results within orders of magnitude for endpoints such as ASPEN angina (2.6% vs. 3.1%); ACCORD-BPLI: heart failure (0.5% vs. 0.8%), major coronary event (3.1% vs. 2.4%), primary outcome (3.6% vs. 2.1%); ACC ORD-GLI non-fatal MI (6.5% vs. 4.6%). CONCLUSIONS: Trial outcomes defined as “new or worsening” were not well-matched by DMAF due possibly to uncertainty in definitions and suitability for modeling.