Abstract：This article proposed a rehabilitative training monitoring system for knee osteoarthritis patients, by which the patient can know their standardized degree of movement and make appropriate adjustments. We designed a quality assessment system of human lower extremity movement which was based on ZigBee wireless communication technology. The system can evaluate the action standardization of movement therapy. We mounted ZigBee module with miniature accelerometer sensor on human lower extremity, by acquiring threedimensional acceleration signal through the movement after Haar wavelet transformation, the system extracts wavelet eigenvalue using particle swarm optimization. These eigenvalues will be supplied in neural network classifier which can evaluate the quality of movement. By evaluating the exercise quality of lying straight leg raise training for 20 healthy men aged 24～30, the mean and standard deviations for these four training truth probability (specification, foot over, hold time is too short, nonparallel leg lifts ) were（89.1±2.0）%，（93.4±1.7）%，（89.5±2.3）%，（90.1±1.8）%. Experimental results show that the system can effectively identify the nonstandard action from training process and realize the monitoring and evaluation of lying straight leg raise training, which can meet the demands of health monitoring system.