Abstract: In this paper dynamic measurement data processing is used to separate the results of measurement, systematic error and the random error.First the method of dynamic data system modeling(DDS ) is applied for preprocessing and preanalysing dynamic measurement data Then various deterministic components are extracted by emphatically applying successive regression analysis and mobified periodogram test or discrete spectral analysis, so as to separate the resutls of measurement and the systematic error. For separating random components their correlation analysis and continuous spectral analysis of the extracted data are conducted by using the DDS mobeling mtheod When necessary it is possible to use the nonlinear least square method to determine better compound models so that accurate estimation can be made for various components mentioned above.The method presented in this paper has the advantage of having detter resolution, higher estimation accuracy and broader applicability. It is expected that by using this method dynamic measurement data prcoessing and systematic error correction can be made on line.