The results of a spectral analysis of a new, subjectively analyzed data set of tropical Pacific wind stress are presented. The monthly data for the 10-year period, 1961–70, allow a detailed inspection of the distributions of frequency and zonal wavenumber spectra from 29°N to 29°S. In addition, the results obtained using the subjective analysis technique are briefly compared with those obtained using two objective methods. The frequency spectra vary greatly throughout the tropical Pacific. There also are differences between the spectra for the wind-stress magnitude and its components. The only statistically significant peaks are for the annual and semiannual cycles. Differences between the frequency spectra for the wind-stress magnitude and the wind-stress components are discussed. Plots of the spatial distributions of the power in the annual and semiannual signals are presented and related to seasonal climatological features in the tropical Pacific wind field. Other plots are introduced which show regions of high interannual variability in the area occupied by the Southern Oscillation, and in the central equatorial Pacific. Both of these regions are key areas in the study of El Niñno. Zonal wavenumber spectra are presented as functions of latitude for January, February, etc. The spectra are red, as would he expected. The outcome of a test for white noise that was performed on the frequency spectra for interannual periods is discussed. According to the data set used in this study, the spectra are indistinguishable from white spectra for interannual periods. However, the results for this type of test are hindered by the short record length available in the data set.