In this article stereoscopically derived cloud top heights and cloud winds estimated from the Multiangle Imaging Spectroradiometer (MISR) are assessed. MISR is one of five instruments on board the NASA Terra satellite. The cloud top height assessment is based on a comparison of more than 4 years of MISR retrievals with that derived from ground-based radar and lidar systems operated by the U.S. Department of Energy Atmospheric Radiation Measurement program. The assessment includes a comparison of the MISR cloud top heights and ground-based data sets as a function of cloud optical depth and a simple cloud classification. Overall, we find that the MISR retrieval is working well with little bias for most cloud types, when the cloud is sufficiently optically thick to be detected. The detection limit is found to be around optical depth 0.3 to 0.5, except over snow and ice surfaces where it is larger. The standard deviation across all clouds is less than about 1000 m for the MISR best winds retrievals at all ARM sites, and the standard deviation for the MISR without winds retrieval varied between about 1000 to 1300 m, depending on the site. The performance for various cloud types is explored.