To reuse 3-D CAD models more efficiently, a 3-D CAD model retrieval algorithm based on distance and angle distributions was proposed. Firstly, a large number of random points on the model surface were taken to calculate the length of the directed line constructed between two random points and the angle between the normal of the point and the directed line. Then, a distance-angle planar grid was constructed to express the distance-angle distribution by obtaining a statistic data of the sampled points. At last, the Manhattan distance metric method was used to compute the similarity between the two distance-angle matrices, which can give the similarity coefficient for two compared 3-D CAD models. Experiments results show that the algorithm can effectively support 3-D CAD model retrieval, and the efficiency meets the requirements of engineering application.