Although the current Multiangle Imaging Spectroradiometer (MISR) and Moderate Resolution Imaging Spectroradiometer (MODIS) satellite passive remote sensing midvisible aerosol optical thickness (AOT) products are accurate overall to about 0.05 or 20%, they differ systematically on a global, monthly average basis, by about 0.03 to 0.05. Some key climate change and other applications require accuracies of 0.03 or better. The instruments are sufficiently stable and well characterized, and have adequate signal-to-noise, to realize such precision. However, assumptions made in the current standard aerosol retrieval algorithms produce AOT biases that must be addressed first. We identify the causes of AOT discrepancies over dark water under typical, relatively low AOT conditions and quantify their magnitudes on the basis of detailed analysis. Examples were selected to highlight key issues for which there are coincident MISR, MODIS, and Aerosol Robotic Network (AERONET) observations. Instrument calibration and sampling differences, assumptions made in the MISR and MODIS standard algorithms about ocean surface boundary conditions, missing particle property or mixture options, and the way reflectances used in the retrievals are selected each contribute significantly to the observed differences under some circumstances. Cloud screening is also identified as a factor, though not fully examined here, as are the relatively rare high-AOT cases over ocean. Specific algorithm upgrades and further studies indicated by these findings are discussed, along with recommendations for effectively using the currently available products for regional and global applications.