This paper describes vehicle speed estimation algorithm that can be seen in rearview mirror. The vehicle speed estimation algorithm consists of preprocessing, vehicle detection, tracking vehicle between frames and speed estimation. Preprocessing ensures higher speed of video frame processing by changing the size of the video frame (if frame is large enough) and only processing part of video frame in which vehicle can be found. The position of vehicle in video frame is gained by detection and tracking the detected vehicle between the frames ensures higher processing rate. The vehicle speed estimation is divided into two parts: distance estimation between vehicle and camera and speed estimation based on prevailing distance between two video frames. The Vehicle speed estimation algorithm has been tested in eleven test video sequences and quite successful results have been achieved, with the space for progress in algorithm work.