Abstract: Automobile manufacturers speed up the launch of new products to meet the changing market demand. A man-hour prediction method for automobile assembly based on assembly similarity and grey model was put forward to solve the problem of slow prediction speed on mixed-model automobile assembly line. Assembly hour was classified into picking, positioning and coupling time by the characteristics, extracting key factors which influence assembly hour for each type. The similarity coefficient between the sample process and benchmark process was computed with factor database, constructing fitting curves in MATLAB combined with working hour data. The positioning and coupling time predictions were gained in the functional expression and GM(0, N) grey model. The accuracy and high-efficiency of proposed method was verified by comparing with Method-Time-Measurement.