OBJECTIVES Previous studies have demonstrated incorporating parameter sampling (PS) is crucial to capture nonlinear effects (NE) in cost effectiveness modeling. NE are, among other causes, driven by the degree through which the symmetric sampling of a risk factor is translated into non-symmetrically distributed probabilities generated by the applied risk equations (RE). This study sought to assess degree by which the incorporation of NE through PS alters event rate predictions from the UKPDS 82 (UK82) and UKPDS 68 (UK68) RE in a set of selected validation studies conducted with the CDM. METHODS A total of 50 validation simulations were performed to data from ACCORD, ADVANCE, VADT, ASPEN, DCCT and UKPDS.  Simulations mirroring cohort baseline characteristics of each of the trials were conducted with and without PS using UK68 and UK82 REs. Predicted versus observed macrovascular (MAC) and microvascular (MIC) complications and all cause mortality (ACM) were assessed using the coefficient of determination (R2) goodness of fit measure. RESULTS When the CDM was run without PS, validation studies produced an R2 statistic of 0.898 using UK68 and 0.853 using UK82 RE.  This compared to R2 statistics of 0.876 and 0.791 in analysis with PS for UK68 and UK 82 REs, respectively. Overall, PS caused end point predictions for MAC, MIC and ACM to increase. Internal validations against UKPDS 80 demonstrated that PS increased event rate predictions for myocardial infarction (MI), stroke, MIC and ACM by 4.4%, 21.5%, 19% and 16.4% when UK68 RE were applied and 26.3%, 64.7%, 14.9% and 34.8% with UK82 RE, respectively. CONCLUSIONS The findings from this study have shown that external validity declined with PS in simulations using UK68 RE and UK82 RE. The degree by which PS increased end point predictions was considerable stronger in UK82 RE predictions for MAC and ACM but lower for MIC.