The Antarctic plateau contains the best sites on earth for many forms of astronomy, but none of the existing bases was selected with astronomy as the primary motivation. In this article, we try to systematically compare the merits of potential observatory sites. We include South Pole, Domes A, C, and F, and also Ridge B (running northeast from Dome A), and what we call “Ridge A” (running southwest from Dome A). Our analysis combines satellite data, published results, and atmospheric models, to compare the boundary layer, weather, aurorae, airglow, precipitable water vapor, thermal sky emission, surface temperature, and the free atmosphere, at each site. We find that all Antarctic sites are likely to be compromised for optical work by airglow and aurorae. Of the sites with existing bases, Dome A is easily the best overall; but we find that Ridge A offers an even better site. We also find that Dome F is a remarkably good site. Dome C is less good as a thermal infrared or terahertz site, but would be able to take advantage of a predicted “OH hole” over Antarctica during spring.