A simple algorithm is presented which identifies cumulus clouds in visible and infrared satellite image pairs. The algorithm is applied to hourly pairs of the Geostationary Operational Environmental Satellite (GOES) to determine the relative frequency of daytime cumulus clouds over the central United States for two consecutive Julys: one drought (1988) and one control (1987). In both Julys, cumulus frequency decreased from southeast to northwest, against the gradient of elevation, but with the gradient of midday average relative humidity. On the scale of uplands and lowlands and even isolated ridges it tended to increase with elevation. During the drought in July, contrasts in vegetative cover were especially large. These contrasts tended to oppose contrasts in surface temperature inferred from satellite radiometers. They also tended to oppose contrasts in the frequency of shallow cumulus. Over central and southern Illinois, a flat to gently undulating, mostly rural plain, during July 1988, shallow cumulus clouds occurred more often over lightly vegetated than heavily vegetated landscapes. These clouds also tended to be relatively abundant near the larger urban centers. During the control July of 1987, fewer cumulus clouds were recorded.