Statistics show that the travel of agricultural machinery to a work area and their movement during labour is the source of many serious accidents. The most dangerous in consequences prove to be those that occur during transport and associated with maneuvering tractors and machinery (about 30% of all fatal accidents). It can be assumed that at least some of these accidents were caused indirectly by the specific design features of agricultural machines which adversely affect the driveability. The single- and multi-loop structures of the driver-vehicle system models are formulated to study the contributions of various preview and prediction strategies to the path tracking and dynamic performance of the articulated vehicle. In the presented study the compensatory model of driver utilizes the lateral acceleration of the tractor, roll angle of trailer sprung mass and the articulation rate as the internal motion feedback variables. The control model of steering of an agricultural set has been implemented in the Matlab/Simulink environment. The model has been constructed with the use of stochastic methods and operational transmittances describing the various components of the system. The model operational transmittances has been estimated using Box-Jenkins and continuous-time process models from input-output data. The model has been tested using experimental data from road investigation of the agricultural set.