The LIMS experiment on NIMBUS 7 used the technique of thermal infrared limb scanning to sound the composition and structure of the upper atmosphere. One of the LIMS channels was spectrally centered at 6.9 μm to measure the vertical profile and global distribution of stratospheric water vapor. This paper describes the characteristics of and data from the water vapor channel and the steps taken to validate results. The mean difference between LIMS measurements and data from 13 balloon under-flights is aout 0.6 ppmv with LIMS mixing ratios biased high with respect to in situ data. This difference is of about the same order as the estimated LIMS accuracy and is less than the sum of the errors for LIMS and the balloon techniques. The precision measured in orbit is 0.2--0.3 ppmv, and the accuracy, based on computer simulations, is estimated to be 20--30% over the range from 50 mbar to about the stratopause. The vertical distribution in the tropics shows the presence of a hygropause where the mixing ratio decreases to a minimum above the tropopause and then increases with altitude. In extratropical regions. The profile is nearly constant with height and has a value of about 5 ppmv. An unexplained diurnal variation exists in the data which is largest at the 1-mbar level (1--2 ppmv) and virtually nonexistant at 10 mbar. Day values ar higher than night. Until this phenomenon is better understood, caution must be used in drawing conclusions aout the H2O altitude behavior in the upper stratosphere. An anti-correlation exists in the fie structure of the vertical water and temperature profiles. This is brought about primarily by small temperature errors and should not be take as real. For this and other reasons, more confidence is placed in zonal mean distributions avaraged over several days rather than in single profiles. A zonal mean pressure-latitude cross section is described for January.