The author comments on the paper of De Roo and F. T. Ulaby (see ibid., vol.42, no.2, p. 220-231, 1994) which described experimental work on scattering by dielectric rough surfaces, and investigated the shift of the Brewster angle when a plane interface becomes randomly modulated. As mentioned by the authors, the numerical evidence of the phenomenon has been given where rigorous computations have shown that the minimum of the reflected intensity is shifted toward lower incidence angles. This conclusion, concerning one-dimensional surfaces under p polarization, is in agreement with the aforementioned experimental work. Later, with the help of the perturbation methods, a more detailed study has been achieved by several authors for both one-dimensional surfaces and two-dimensional surfaces. The main point is that although the theories are not the same, they all lead to the same conclusion and confirm the previous results. Therefore, the author was very surprised to read in the introduction that a shift toward grazing angles was predicted. The DE Roo and Ulaby reply that the theoretical and numerical calculations predict that the Brewster angle shifts in the direction of normal incidence (negative shift) as the surface roughness increases, and their experimental results indeed support the prediction. They have no disagreement with this comment and are in fact pleased that their data confirms the theory. >