To address the problems of insufficient differentiation in diagnosability evaluation and the difficulty in interpreting the evaluation results,a quantitative diagnosability evaluation method based on Wasserstein metric is proposed.The method is based on the measured signals from the measurement points of the system in different states,regards the signals as random variables and uses the 2-Wasserstein metric as a measure of the degree of similarity between the variables to achieve the evaluation of the diagnosability of the system.And an alysis is made on the impact of the measured signals on the results of the diagnosability evaluation.The method proposed is validated by using a double bandpass filter circuit,and the results show that compared with the diagnostic evaluation method based on K-L divergence and D matrix,the method in this paper is more objective and scientific in evaluating fault diagnosability.