On-line dynamic balancing is the main method to solve the spindle system unbalance failure and reduce the spindle vibration. In order to improve the balancing accuracy and efficiency of the on-line dynamic balancing device， an optimization model of the spindle system dynamic balancing quality compensation strategy based on genetic algorithm is proposed. The model takes the residual unbalanced force as the optimization objective and the rotation phase of the mass block of the balancing device as the optimization variable. Then the phase is calculated by genetic algorithm. The on-line dynamic balance test platform of the spindle system is built to carry out the comparison and optimization experiment of the dynamic balance strategy under different rotating speeds， and the residual unbalanced force after the balance is greatly reduced. The experimental results show that after the optimization of the mass compensation strategy， the vibration of the spindle decreases by an average of 20.60%， and the balance time decreases by an average of 34.67%. It can be seen that the optimization model of the quality compensation strategy of the spindle system based on genetic algorithm can further improve the quality and efficiency of the spindle on-line dynamic balance and improve the operating performance of the equipment.