Based on the performance of magnetorheological damper we have developed,a tremor suppression method using human limb nonrigid characteristic was presented.The tremor suppression machine was designed,and the measure part and the tremor suppression part were separated.Subsequently,the mathematic models of fuzzy neural network and elbow nonrigid tremor were worked out.Using selected training data set,the off-line fuzzy neural network training was finished.Then the tremor suppression method was simulated in Matlab.From the simulation results,it is found that the tremor suppression method has presented suppresses tremor effectively,and the tremor amplitude and tremor suppression torque can be controlled with angle displacement sensor.