OBJECTIVES: The Cardiff Type-2 Diabetes Model is a fixed time increment stochastic simulation model written in Microsoft Excel and C++; initially developed in 2003, it has been used to support a number of cost-effectiveness and public health policy decisions.  The model is fundamentally built around the risk equations reported from the United Kingdom Prospective Diabetes Study (UKPDS) and concerns persist regarding the validity of these equations in contemporary populations. Therefore, the objective of this study was to validate the model’s output to recently published outcomes trials to establish if the UKPDS equations remain credible. METHODS: Simulated cohorts reflecting the baseline characteristics associated with key outcomes studies (ASPEN, ADVANCE, ACCORD, VADT, ADDITION, ASCOT, CARDS and long term follow-up of UKPDS) were generated and treatment effects applied to reflect intensive versus conventional arms.  Predicted and observed events, over a time horizon consistent with each trial, were compared and goodness of fit determined, using the coefficient of determination (R2). RESULTS: Across all validation studies predicted versus observed events resulted in an R2 statistic of 0.90. This result was obtained when including data from UKPDS, for which the model gave an exceptionally fit (R2 = 0.95) When excluding UKPDS the overall R2= 0.7. Despite the less accurate fit, there was a consistent trend demonstrated from the model although a noteworthy lack of fit was observed for the ACCORD blood pressure trial (non-fatal myocardial infarctions [MI]) and ACCORD glucose lowering trial (non- fatal MI and congestive heart failure) in which the predicted events rates by the model were substantially lower than reported in the trials. Similarly, for the ADVANCE trial, non-fatal strokes were significantly under-predicted by the model. CONCLUSIONS: This study suggests the UKPDS risk equations within type-2 diabetes models remain credible for supporting contemporary technology assessment and health policy decisions.