This study developed a testing system for fabric surface softness based on the sled method. The system is compact, lightweight, portable and convenient to use. The device can be installed on the sueding machine as an on-line real-time testing system for fabric surface softness, thus improving the response efficiency of the process parameters of the sueding machine. Moreover, it can be used to construct a database of these parameters, store different fabric types, process quality requirements and reproduce processing at the same quality, in order to automatically, rapidly and accurately calibrate the parameters. This study applied the Taguchi method on the process parameters of the sueding machine to analyze the interaction between parameters, design the orthogonal array according to actual needs, and process the grinded fabric on-line. It also measured the testing system and plotted the Signal-to-Noise (S/N) ratio response table based on the S/N calculated for the parameter combinations. In the table, the larger S/N ratios represent a better quality characteristic and smaller variation. Hence, the main effect of the control factors and the optimum factor level can be obtained. After determining the optimal combination of process parameters, the process and measuring experiment of the reference group was conducted with these control factors to verify the results. It was proven that the device can obtain the maximum surface softness, and effectively and accurately test the surface softness of finished fabrics.