Current spectral analysis for evaluating the rail ride quality effectively is based on the sampling data at a uniform time interval, but the train is of fluctuation velocity in motion, which results in a non-uniform interval between consecutive sampling data. Therefore the accuracy of routine spectral analysis is in doubt when applying it in evaluating the rail ride quality. This paper presents a new way, namely, re-sampling with variable frequency to eliminate the influence of the train’s uneven velocity. Its feature is that there is no need for precision measurement of the train’s moving speed. Experiment results from the test-bed of rolling stock vibration show that it is valid. Current spectral analysis for evaluating the ride ride quality effectively based on the sampling data at a uniform time interval, but the train is of fluctuation velocity in motion, which results in a non-uniform interval between consecutive sampling data. spectral analysis is in doubt when applying it in evaluating the rail ride quality. This paper presents a new way, namely, re-sampling with variable frequency to eliminate the influence of the train the uneven velocity. Its feature is that there is no need for precision measurement of the train’s moving speed. Experiment results from the test-bed of rolling stock vibration show that it is valid.