Abstract: Considering the influence of different values of collected data on health status for radar transmitter, the method of weighting sample data with exponential function is proposed, and then the health status evaluation model is constructed with support vector machine (SVM) as the discrimination algorithm. This method can effectively highlight the influence of individual state data deterioration on the whole health state by adjusting the weight factor. In the case of a small number of sample training, it can significantly improve the accuracy of health assessment. The practical test shows that this method has practical value in the accurate evaluation of radar transmitter health.