Registration of range images plays a key role in 3D digital imaging and modeling and 3D optical inspection. Its precision and velocity directly determine the quality of 3D model construction. At present, two registration methods, iterative closest points and markers, are in wide application. All kinds of methods for range images registration are summarized, where iterative closest points method and markers method are introduced mainly. According to experimental results, these two methods are compared in various regions, including registration-error, registration-time, error-source, application-areas and so on.