This study investigated the consistency between microwave and optical water path estimates of oceanic clouds from Tropical Rainfall Measurement Mission (TRMM) Microwave Imager (TMI), Moderate Resolution Imaging Spectroradiometer (MODIS), and Multiangle Imaging Spectroradiometer (MISR). We used microwave estimates from the Wentz algorithm for warm, nonprecipitating clouds and both Wentz retrievals and standard TMI profiles for cold, precipitating clouds. Optical estimates were derived from cloud optical thickness and particle effective radius. For warm, nonprecipitating clouds the two methods showed good agreement at the 25-km resolution of the microwave measurements, with liquid water path means being within 5--10%, an overall correlation of 0.85, and RMS difference of ~25 g m-2. Multiangle optical retrievals showed only weak variations with view zenith angle, building further confidence in the results. An error analysis suggested that optical estimates were more certain than microwave ones, primarily because 1-D plane-parallel radiative transfer seemed to apply well at the coarser comparison scales. However, there appeared to be a slight but systematic dependence on cloud amount as microwave retrievals increasingly overestimated optical ones at cloud fractions below ~65%. For cold, precipitating clouds we tested three common interpretations of optical retrievals: liquid water path, ice water path, and total water path. The relationship between microwave and optical estimates was weak in all cases, with correlations no more than 0.5, and RMS differences at least an order of magnitude larger than for warm clouds. The weakest correlation (0.37) was found when optical retrievals were interpreted as ice water paths. If anything, optical estimates appeared best correlated with microwave cloud liquid water paths.