Data from spaced-antenna experiments are normally analysed by correlation techniques in the time domain, but several authors have recently pointed out that an equivalent analysis can be carried out in the frequency domain. To explore any possible advantages of such a procedure, we assume Gaussian forms for the correlation functions so that their Fourier transforms can be evaluated analytically. The power spectrum is shown to be a convolution of two terms, one of which represents “beam broadening” due to the mean wind, and the other represents the broadening due to turbulent motions. The cross spectrum has a phase which is a linear function of frequency, the slope of which is related to the “apparent velocity” of correlation analysis. If this same cross-spectral phase is interpreted by the method of Doppler interferometry, the velocity obtained is also the apparent velocity. The Gaussian model is extended to include the effects of vertical velocity and of scatter which is centered off zenith. A computational scheme is presented in which an analysis equivalent to conventional “full correlation analysis” is carried out entirely in the frequency domain. The method is illustrated using data from Adelaide.