Abstract: To verify the practical applicability and accuracy of the large mass method (LMM) widely used in multiple-support seismic excitation analysis of structures, numerical simulations of a two-degrees-of-freedom (2-DOF) finite element model adopting the Rayleigh damping assumption are performed using the LMM and the relative motion method (RMM) respectively under four groups of seismic ground motions. Through comparisons with the RMM, the origin of errors and the applicability of the LMM are discussed. Then the improved LMM is presented herein based on the modification of ground motions considering the influences of Rayleigh damping coefficient α. It indicates that the LMM is not applicable to multi-support excitation analysis in the case of Rayleigh damping, which can cause significant errors. And the errors depend on the damping coefficient α. It is also proved that the improved LMM is able to yield results that are identical to those of the RMM.