In order to detect missile targets as early as possible, atmospheric absorption bands are used in the U.S Defense Support Program (DSP). To explore the band settings and their selection methods for the DSP,a new method to choose the optimal detection bands based on a background clutter model is proposed. The simplified parametric expression for a scene radiation is deduced, and then the spatial clutter model is established. The main factors impacting the clutter radiance are analyzed in absorption infrared wavebands,and the Synthetic Signal-to-noise Ratio(SSNR) is given in an analytic form. Finally, the optimal bands are chosen in all the possible scene clutter levels and target radiation characteristics, which provides the best system performance for detecting the same kinds of targets with the similar radiation characteristics under different clutter levels. The validity of the method is illustrated by examples. Calculation and analysis results indicate that the optimal detection bands are dependent on the scene clutter level and target radiation characteristics. For detection two different engine plumes, the optimal bands for DSP are 2.73-2.85 μm and 4.2-4.43 μm.