Gear trains applied to automotive transmissions and combustion engines are potential excitation sources of undesired whine noise. Consequently, the prediction of gear whine issues in an early stage of the product development process is strongly requested. Beside the actual excitation mechanism which is closely related to the gear's transmission error, the vibratory behavior (e.g. resonances) of other affected components like shafts, bearings and housing plays an important role in terms of structure borne noise transfer. The paper deals with gear contact models of different degree of detail, which are embedded in a multi-body dynamics (MBD) environment. Since gear meshing frequency and their harmonics may easily reach up to 5 kHz or even 10 kHz, applied gear contact models must be highly efficient with respect to calculation performance. Otherwise, major requirements of the development process in terms of process time can not be satisfied as is the case with FEA-based contact models. Models are developed for spur and helical gears, containing contact point search and force calculation. Tooth deformations are covered as well as friction forces. Resolution of contact is preliminary based on the ideal tooth geometry, deviations from the ideal geometry caused by shape modifications and manufacturing tolerances are considered in a subsequent step. In order to cover edge loading effects the approach reflects gear misalignment, too. A comparison to experimental data for a two gear stage test-rig is performed across the engine speed range. It turned out that the knowledge of the detailed flank geometry is the major prerequisite for a reliable prediction of gear whine excitation and transmission error.