In this paper, a novel hybrid method for path planning problem of multiple mobile sensors on a 3-D terrain is proposed. Our method proceeds in two phases: the global path-planning phase, and the local path planning phase. The first phase constructs a connectivity graph generated by a probabilistic roadmap (PRM) method and selects the control points of sensors' paths from the set of nodes generated by the PRM method. In the local path-planning phase, a hybrid evolutionary algorithm is proposed to determine the intermediate points, which are between control points of sensors' paths in order to complete the paths. The local-path planner considers the accessibility of control points, smoothness of each path, visibility of terrain covered by mobile sensors and the total cost of all paths (i.e. the total length of all paths). The experimental study points out the effectiveness of our framework under various terrain and sensor characteristics.