[en] Assuming that earthquake ground motion is a stationary time function, the seismic analysis of a linear structure can be done by probabilistic methods using the 'power spectral density function' (PSD), instead of applying the more traditional time-step-integration using earthquake time histories (TH). A given structure was analyzed both by PSD and TH methods, computing and comparing 'floor response spectra'. For the purpose of having 'equivalent' TH and PSD to use in the comparison, both were derived artificially from the USNRC-response spectra. The TH were readily available; the PSD was derived directly from the USNRC-response spectra by iteration. The analysis using TH was performed for two different TH and different frequency intervals for the 'floor-response-spectra'. The analysis using PSD first produced PSD functions of the responses of the floors and these were then converted into 'floor-response-spectra'. Plots of the resulting 'floor-response-spectra' show: (1) The agreement of TH and PSD results is quite close. (2) The curves produced by PSD are much smoother than those produced by TH and mostly form an envelope of the latter. (3) The curves produced by TH are quite jagged with the location and magnitude of the peaks depending on the choice of frequencies at which the 'floor-response-spectra' were evaluated and on the choice of TH