OBJECTIVES: Updated risk equations are available for predicting outcome in people with type 2 diabetes (T2D): the UKPDS Outcomes Model (UKPDS 68). It is important to assess the validity of applying risk equations to populations other than those from which they were derived. The objective was to evaluate how well the UKPDS-68 equations predicted vascular morbidity and mortality in real-life data from Cardiff, UK, and compare estimates with the previous UKPDS Risk Engine equations (UKPDS-RE [from UKPDS publications 56 and 60]). METHODS: The equations were incorporated into a stochastic simulation model that estimated the incidence and prevalence of complications (DiabForecaster). Predicted results from the model were compared with population data from Cardiff for coronary heart disease (CHD), stroke and all cause mortality. The annual incidence of newly diagnosed T2D, baseline modifiable risk factors and demographic profiles were matched to the Cardiff data. RESULTS: Internal validation, using a baseline cohort matched to the UKPDS study, demonstrated that the model predicted 12-year cumulative incidence in line with previous UKPDS publications. Real life and predicted event rates for CHD were: 116, 153 and 137 events/1000 T2D patients/yr for the Cardiff data, UKPDS-RE and UKPDS-68, respectively. For stroke: 178, 153 and 128 events/1000 T2D patients/yr, respectively. For all cause mortality: 418, 430, and 475 events/1000 T2D patients/yr, respectively. CONCLUSIONS: All UKPDS equations demonstrated internal validity when compared with published UKPDS data, however both UKPDS-RE and UKPDS-68 equations over predicted the incidence of CHD and mortality and under predicted stroke. While all endpoints predicted were reasonably concordant with observational data discrepancies between UKPDS-68 and UKPDS-RE are worthy of further investigation.