Most daily driving tasks are of low bandwidth and therefore the relatively slow visual system receives enough cue information to perform the task in a manner that is statistically indistinguishable from reality. On the other hand, evasive maneuvers are of such a high bandwidth that waiting for the visual cues to change is too slow and skilled drivers use steering torques and vestibular motion cues to know how the car is responding in order to make rapid corrective actions. In this study we show for evasive maneuvers on snow and ice, for which we have real world data from skilled test drivers, that the choice of motion cuing algorithm (MCA) settings has a tremendous impact on the saliency of motion cues and their similarity with reality. We demonstrate this by introducing a novel optimization scheme to optimize the classic MCA in the context of an MCA-Simulator-Driver triplet of constraints. We incorporate the following four elements to tune the MCA for a particular maneuver: 1) acceleration profiles of the maneuver observed in reality, 2) vestibular motion perception model, 3) motion envelope constraints of the simulator, and 4) a set of heuristics extracted from the literature about human motion perception (i.e. coherence zones). Including these elements in the tuning process, notwithstanding the easiness of the tuning process, respects motion platform constraints and considers human perception. Moreover the inevitable phase and gain errors arising as a major consequence of MCA are always kept within the human coherence zones, and subsequently are not perceptible as false cues. It is expected that this approach to MCA tuning will increase the transfer of training from simulator to reality for evasive driving maneuvers where students need training most and are most dangerous to perform in reality.