A two-dimensional histogram method for cloud analysis from NOAA satellite high-resolution imagery is presented. The algorithm, using one visible or near-infrared channel and one infrared channel of the Advanced Very High Resolution Radiometer (AVHRR), applies to arrays segmented according to the temperature sounder (TOVS) grid; up to five cloud levels are identified in addition to the surface. A comparison with visual nephanalysis has been carried out for various cloud situations: the use of near-infrared radiance appears to give consistently better results than the visible radiance, leading to an agreement in about 87% of the cases for total cloud cover. Possible further improvements are discussed.