We have developed a novel approach for the simultaneous detection and separation of mineral dust from cirrus clouds over major dust outbreak and transport areas using the MODIS thermal infrared window brightness temperature data. Based on the spectral variability of dust emissivity at the MODIS 3.75, 8.6, 11 and 12 μm wavelengths, we integrated heritage threshold tests, including D*-parameter, BTD-slope and BTD3-11 approaches, to identify dust and cirrus pixels. To demonstrate the effectiveness of this detection/separation method, MODIS data for three dust-laden scenes have been processed and analyzed. The detected daytime dust and cloud coverage for the scene covering Persian Gulf compare reasonably well to those from the “Deep Blue” solar retrieval algorithm developed at NASA-GSFC. The nighttime dust and cloud detection for scenes containing the Cape Verde Islands and Niger, West Africa have been validated by comparing with collocated and coincident ground-based micro-pulse lidar measurements. This technique can be further modified using corresponding NPOESS-VIIRS thermal infrared window channel data for dust and cloud remote sensing.