For the features of the modern urban road network that there are complex roads, range of intersections and the main roads and side roads are parallel to each other, map-matching algorithms of vehicle navigation are studied in this paper, and a new map-matching algorithm is proposed. The new algorithm includes candidate roads identification based on computational geometry and GPS (Global Positioning System) points matching based on distance, head, road connection and turn restriction weights. The application in the road network of Beijing shows that the algorithm has high efficiency and accuracy, and is capable of meeting the requirements of map-matching of vehicle navigation systems.