Abstract. Finding single pair shortest paths on surface is a fundamental problem in various domains, like Geographic Information Systems (GIS) 3D applications, robotic path planning system, and surface nearest neighbor query in spatial database, etc. Currently, to solve the problem, existing algorithms must traverse the entire polyhedral surface. With the rapid advance in areas like Global Positioning System (GPS), Computer Aided Design (CAD) systems and laser range scanner, surface models are becoming more and more complex. It is not uncommon that a surface model contains millions of polygons. The single pair shortest path problem is getting harder and harder to solve. Based on the observation that the single pair shortest path is in the locality, we investigate in this paper efficient methods by excluding part of the surface model without considering them in the search process. Three novel expansion-based algorithms are proposed, namely, Na¨ıve algorithm, Rectangle-based Algorithm ...