Computer-based models of two heavy vehicles were created by two consultants using three separate computer-based modelling packages: ADAMS, UMTRI's constant velocity Yaw/Roll program and AUTOSIM. The same input datasets were provided to each consultant and simulations were performed using the same test manoeuvres. A wide range of variables from the simulations was compared. Simulations involving only vehicle responses were found to be in very close agreement. Simulations involving a driver model were generally found to be in good agreement. Recommendations are made that will reduce variability between models in path following tasks to acceptable levels. A generalised dataset fully describing the two vehicles is presented that would be suitable for comparing computer-based modelling packages and for initial testing of potential service providers. Also, a review was performed of studies that have verified predictions from heavy vehicle simulations against test data. (a)