We present a new preoperative planning method for reducing the risk associated with the insertion of straight tools in image-guided keyhole neurosurgery. The method quantifies the risks of multiple candidate trajectories and presents them on a physical model of a head using Augmented Reality (AR) to assist the neurosurgeon in selecting the safest path. The surgeon can then define and/or revise the trajectory in the physical space with AR visualization of risk structures such as blood vessels and ventricles, tool placement uncertainty, and quantitative risk measurements. Then, the neurosurgeon can revise the selected path on the 2D MRI image slices to incorporate all relevant information. Finally, a simulation of the surgery can be performed on the physical head model for a more detailed examination of the possible risks. Our preliminary results on clinical data show that in complex situations the method can improve risk assessment.