Background: The structure of brainstem is very complex, and the surgical path planning for surgical navigation system can reduce the damage of the important tissue. We proposed an automatic preoperative path planning method based on combined magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI) for determining optimal surgical paths in brain stem tumor surgery. Methods: First, we register the DTI into the MRI by comprehensive considering three type's information of trace, fractional anisotropy and relative anisotropy to achieve more accurate DTI location information. After that, the optimal solution model of the preoperative path which computes a cost function associated with each point on the outer brain boundary and instrument entry path, is constructed by the gravitational repulsion model and spline interpolation using the segmented model of brainstem and fiber bundle. Furthermore, the preoperative path can be achieved automatically by optimizing the solution of the cost function, and the results are evaluated by comparing the cost of a particular path associated with each critical structure, as well as the total number of examining all the cross-sectional images orthogonal to this path. Results: Our method could complete automatic preoperative path planning and avoid important tissue damage with less cross-sectional images orthogonal to the planning path.