Desert dust has a potentially strong effect on climate and climate forcing as evident from satellite data, yet its detection close to its sources over deserts and semidesert regions is difficult. The lifted dust often has the same mineral composition as the surface sand left behind. The main consistent difference between airborne dust and surface sand is particle size. This difference is used here to develop a remote sensing technique for the detection of dust over desert. The technique uses two infrared channels that will be on the Earth Observing System (EOS) moderate resolution imaging spectroradiometer (MODIS). It is based on measurements and modeling of the thermal infrared spectrum of silicate and carbonate minerals that show strong differences between the thermal properties of the smaller dust particles (diameter of 2--5 μm) and the larger sand particles (diameter >70 μm). In particular, the emissivity of quartz increases with the particle size at 11 μm and decreases at 8.6 μm. This different behavior allows use of these two MODIS channels to distinguish between variations of the dust temperature and surface temperature from the variation in the loading of the smaller dust particles in the atmosphere. We report simulations of the technique for day and night remote sensing and apply it to high resolution infrared sounder (HIRS) data which is validated against ground-based measurements by Sun photometers. The technique is found to be sensitive to the dust loading over sand surfaces and not very sensitive to the dust height. ¿ 1998 American Geophysical Union