Surface scattering effects are merely diffraction phenomena resulting from random phase variations induced on the reflected wavefront by microtopographic surface features. The Rayleigh-Rice and Beckmann-Kirchhoff theories are commonly used to predict surface scattering behavior. However, the Rayleigh-Rice vector perturbation theory is limited to smooth surfaces, and the classical Beckmann-Kirchhoff theory contains a paraxial assumption that confines its applicability to small incident and scattering angles. The recent development of a linear systems formulation of nonparaxial scalar diffraction phenomena, indicating that diffracted radiance is a fundamental quantity predicted by scalar diffraction theory, has led to a reexamination of the classical Beckmann-Kirchhoff scattering theory. We demonstrate an empirically modified Beckmann-Kirchhoff scattering model that accurately predicts nonintuitive experimental scattering data for rough surfaces at large incident and large scattering angles, yet also agrees with Rayleigh-Rice predictions within their domain of applicability for smooth surfaces.