In support of its dual mission in environmental studies and nuclear nonproliferation, the Multispectral Thermal Imager (MTI) has enhanced spatial and radiometric resolutions and state-of-the-art calibration capabilities. These instrumental developments put a new burden on retrieval algorithm developers to pass this accuracy on to the inferred geophysical parameters. In particular, current atmospheric correction schemes assume the intervening atmosphere is adequately modeled as a plane-parallel horizontally-homogeneous medium. A single dense-enough cloud in view of the ground target can easily offset reality from the calculations, hence the need for a reliable cloud-masking algorithm. Pixel-scale cloud detection relies on the simple facts that clouds are generally whiter, brighter, and colder than the ground below; spatially, dense clouds are generally large, by some standard. This is a good basis for searching multispectral datacubes for cloud signatures. However, the resulting cloud mask can be very sensitive to the choice of thresholds in whiteness, brightness, and temperature as well as spatial resolution. In view of the nature of MTI’s mission, a false positive is preferable to a miss and this helps the threshold setting. We have used the outcome of a genetic algorithm trained on several (MODIS Airborne Simulator-based) simulated MTI images to refine an operational cloud-mask. Its performance will be compared to EOS/Terra cloud mask algorithms.