The direct use of measured radiances for determining the thickness of stratospheric layers is investigated. We hypothesize that the equivalent blackbody temperature, weighted according to the transmittance weighting functions for the stratospheric channels of the satellite infrared spectrometers and the selective chopper radiometer, gives a good approximation of the geometric mean temperature of some layer within the transmittance (τv) domain 0<τv <1. A priori, it is shown that under certain conditions this is not a good assumption. However, it is of interest to determine for what atmospheric layers acceptably small error in the mean temperature, and therefore in the thickness, would be incurred. Layers based at 100-10 mb, with upper boundaries at 10-0.5 mb, are investigated using a carefully selected family of stratospheric temperature profiles and computed radiances. On the basis of physical reasoning, a high correlation of thickness with radiance is anticipated for deep layers, such as the 100- to 2-mb layer (from about 15 to 43 km), that emit a substantial part of the infrared energy reaching a satellite radiometer in a particular channel. Empirical regression curves relating thickness and radiance are developed and are compared with “blackbody” curves obtained by substituting the blackbody temperature in the hydrostatic equation. Maximum thickness-radiance correlation is found, for each infrared channel, for the layer having the best agreement of empirical and blackbody curves. For these layers, the data from a single radiation channel accounts for a reduction of variance by up to 97 percent. The utility of thickness data based on actual radiances is demonstrated through independent testing and with a sample 2-mb map constructed by adding thicknesses based on measured radiances to the observed 100-mb height field.