The operation status of radar power supply system directly affects the safety and performance of the whole radar equipment. How to realize the health status assessment of radar power supply is a problem to be solved. In this paper, a health assessment model is established to achieve this goal. Firstly, the health feature data of radar power supply under normal state is used to train the self-organizing neural network. Then the distance between the monitoring data and the weight vector of the trained self-organizing neural network is calculated. The distance value is converted into the health degree between (0,1). Finally, the health degree is calculated by using the data of accelerated degradation test, According to the calculation of health degree, the health grade is divided. The results show that the health degree calculated by the proposed method decreases with the aging of power supply, which indicates that the method can be used to evaluate the health status of radar power supply.