Roughness, reflecting the smooth degree of the surface, is a key factor of many land surface processes. There are various methods of parameterization, but questions still exist since the characterization of land surface system has far from been fully understood. This paper reviews the research progress on land surface roughness characterization from three aspects, including land surface measurement techniques, roughness parameters, and remote sensing researches. So far, pin profiler and laser profiler are still main-stream land surface measurement approaches, while 3D laser scan and photogrammetry techniques have shown their potentials. Parameters defined by statistical methods and fractal theories are quite distinct from each other, but it is hard to describe land surface using either category of parameters due to the the complexity of its multi-scaled feature. As for remote sensing researches, both optical and microwave techniques have broad prospects, and the former needs to do better in match with classical parameterization, while the latter has to make improvement on techniques and methods for next generation remote sensing platform. Many key problems encountered in roughness parameterization study, such as comparison and transformation of parameters between different scales, heterogeneity and anisotropy of land surface, as well as roughness parameterization based on 3D surface data, have also been discussed.