Clear-conservative satellite cloud detection algorithms overestimate cloud fraction (CF), the degree to which depends on instrument resolution and the spatial distribution of cloud area, with boundary layer cumulus carrying the largest overestimation. This is because in the standard method of computing CF, partially cloudy pixels contribute in the amount of the pixel’s area rather than the true cloud area. Development of analytical and pattern recognition techniques to reduce overestimation has been limited by the lack of coincident long-term, large scale, and high-resolution satellite data sets. Such data sets are now available from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and Multiangle Imaging SpectroRadiometer (MISR) instruments onboard the satellite, Terra. These data sets are used to determine the resolution required to reduce the CF overestimation to ≤0.01 absolute CF using the standard, analytical, and pattern recognition techniques on perfect, clear-conservative cloud masks with future instrument design in mind; investigate the challenges of implementing these techniques on MISR’s operational clear-conservative cloud mask; and demonstrate the impact of these techniques on a MISR-derived CF climatology similar to those used to evaluate climate models. Reducing the median CF bias to ≤0.01 requires resolutions of <45 m and ≤80 m for the standard and analytical techniques, respectively, while the pattern recognition technique has no resolution requirement up to a resolution of 1.2 km. When the pattern recognition technique is applied to 10-year, December, low cloud climatologies over the tropical Western Atlantic (10 N–20 N and 50 W–60 W) derived from MISR, the average CF reduces from 0.50 to 0.20. This is a large reduction in cloud fraction, particularly when considering the climate sensitivity to low clouds and the use of satellite derived CF climatologies for evaluation of climate model CF and radiative budgets.