The depletion systems defined by the general purpose evaluated nuclear data libraries are unnecessarily complex for most applications in nuclear reactor physics analysis. However, the corresponding compression methods are confronted with two difficulties. On one hand, the number of possible compressed depletion systems is excessively large. On the other hand, the complicated neutronic-depletion coupling effects should be properly considered. In spite of the legacy empirical-based or semi-empirical-based methods, a generalized depletion system compression method based on quantitative significance analysis is proposed in this paper. First, a quantitative significance pair was defined for each basic unit compression operation (BUCO) with respect to the neutron production density, neutron absorption density, and number densities of selected important nuclides. Second, a series of representative problems was composed according to the problem definition domain and simulated by using the original depletion system. Third, the significance pairs were evaluated based on the simulation results of the representative problems, and then employed as the quantitative guidance for accepting or rejecting each BUCO. The commpressed depletion systems have been obtained based on the newly proposed method, and typical pressurized water reactor problems were employed to verify the compresssed depletion systems. Numerical results demonstrated that by adopting the compressed depletion systems generated by the proposed method, significant computing time and storage savings can be achieved while maintaining demanded accuracy.