Near-nadir, quasi-specular backscatter data obtained with a 14-GHz airborne radar altimeter are analyzed in terms of the surface mean square slope (mss) parameter. The raw mss data derived from a least squares fitting of a ray optical scattering model to the return waveform, show an approximately linear wind speed dependence over the wind speed ranges of 7--15 m s-1, with a slope of about one half that of the optically determined mss. Futher analysis based on a simple two-scale scattering model indicates that, at the higher wind speeds, ~20% of this apparent slope signal can be attributed to diffraction from waves shorter than the estimated diffraction limit of ~0.10 m. The present slope data as well as slope and other data from a variety of sources, are used to draw inferences on the structure of the high wavenumber portion of the wave spectrum. The data support a directionally integrated model height spectrum consisting of wind speed dependent k-5/2 and classical Phillips' k-3 power law subranges in the range of gravity waves, with a transition between the two subranges occurring around 10 times the peak wavenumber, and a Durden and Vesecky wind speed dependent spectrum in the gravity-capillary wave range. With a normal value of the spectral constant Au=0.002 in the first k-5/2 subrange, this equilibrium spectrum model predicts a mss wind speed depedence that accords with much of the available data at both microwave and optical frequencies. ¿American Geophysical Union 1992