A numerical evaluation and analysis of Bahar's extended full wave theory for rough surface scattering is carried out. It is shown that the field correlation function in the extended full wave theory is dramatically different from that in the Kirchhoff theory. For this reason the approximation that the heights and slopes are uncorrelated is not valid. When the statistical analysis is carried out without any approximation no enhanced backscatter effects appear. Various artifacts in the theory, such as the pole in the scattering coefficient for vertical polarized waves incident on a rough surface, are also discussed. The revised version of the full wave theory, introduced by Bahar in 1992, is also examined. It is shown that this version of the theory reduces to the small perturbation theory in the small height small slope limit but only for a restricted range of surface parameters.This latest version of the theory eliminates the troublesome pole, agrees with the Kirchhoff theory in the high frequency small slope limit, but produces an unnatural enhancement of the cross sections in the quasi specular direction for surfaces with large rms slopes.