By varying the amount of water vapor as input to the radiative power transfer equation, assuming a constant carbon dioxide and varying ozone distribution, it is possible to infer stratospheric water vapor from broadband observations of downward irradiance. The procedure is iterative in that downward observed and calculated irradiances, at several levels for each of several radiometric soundings, are brought within the limits of a convergence criterion. This is accomplished by successively reducing an initial over-estimate of the stratospheric mixing ratio, defined by a power law, until the sum of the squared differences of observed and calculated irradiances is minimized. The sum includes all levels of the sounding. Results for a continental area during winter months indicate that the stratospheric water vapor content from 50 mb upward to 10 mb decreases from approximately 20 to 3 parts per million. For tropical Guam and Canton Island the corresponding magnitudes are larger, decreasing from 21 to 4 ppm. The standard deviation of the mean for all pressure levels is approximately 1.0 ppm. Adding deviation to the values inferred should give an upper bound to the water vapor content. The average mixing ratio for the continental stations between 25 and 10 mb is 5.7 ppm with a standard deviation of the mean of 0.8 ppm. Since the infrared radiative emission and attenuation of aerosols is inseparable from emission and attenuation of the atmospheric gases when measured with a broad response radiometer, these mixing ratio results would be reduced by the presence of aerosols. In view of apparent aerosol contamination we have made no inferences below 50 mb (21 km). The results may be said to be an upper bound to the actual quantity of water vapor, favoring an increasingly dry stratospheric profile.