The highlands of Venus are characterized by an altitude-dependent change in radar backscatter and microwave emissivity, likely produced by surface-atmosphere weathering reactions. We analyzed Magellan and Arecibo data for these regions to study the roughness of the surface, lower radar-backscatter areas at the highest elevations, and possible causes for areas of anomalous behavior in Maxwell Montes. Arecibo data show that circular and linear radar polarization ratios rise with decreasing emissivity and increasing Fresnel reflectivity, supporting the hypothesis that surface scattering dominates the return from the highlands. The maximum values of these polarization ratios are consistent with a significant component of multiple-bounce scattering. We calibrated the Arecibo backscatter values using areas of overlap with Magellan coverage, and found that the echo at high incidence angles (up to 70¿) from the highlands is lower than expected for a predominantly diffuse scattering regime. This behavior may be due to geometric effects in multiple scattering from surface rocks, but further modeling is required. Areas of lower radar backscatter above an upper critical elevation are found to be generally consistent across the equatorial highlands, with the shift in microwave properties occurring over as little as 500 m of elevation. These surfaces are not simply characterized, however, by the absence of a highly reflective component. Surface morphology and radar-scattering properties suggest that a mantling deposit forms at the highest elevations, most likely by in situ erosion of the original rock. In Ovda Regio, this process mantles or has removed surface festoon structure at the 1- to 10-m scale, implying a significant depth for the weathered layer. Similar radar-dark areas occur in Maxwell Montes but are apparently unrelated to the current topography of the region. Possible reasons for these observations include mass wasting from areas of steep slopes, compositional or age differences within the montes, vertical tectonic shifts of relict contacts, local topographic effects on surface temperature, or errors in the Magellan topography data in the rugged terrain. While there is evidence for some of these effects in the existing data, no single model at present appears to satisfy all occurrences of high-altitude, radar-dark terrain. New measurements of the surface and lower atmosphere chemistry of Venus are needed to further refine these conclusions. ¿ 1999 American Geophysical Union