This paper presents an analysis of upper tropospheric relative humidity and clouds determined from geostationary satellite observations of the upwelling infrared radiation. The 6.7-mum channel is located near the center of a strong water vapor absorption band and under clear sky conditions is primarily sensitive to the relative humidity averaged over a depth of atmosphere extending from 200 to 500 mbar. Estimates of the clear sky radiance at 6.7-mum are obtained by utilizing the local spatial structure of the radiance field at 11-mum and the correlation between ll-mum and 6.7-mum, to discriminate between clear and cloudy pixels. This approach is demonstrated to be more reliable than cloud clearance based solely upon the local spatial structure of the 6.7-mum channel alone and provides estimates of the clear sky 6.7-mum brightness temperature for areas 2-degrees x 2-degrees of latitude and longitude which are repeatable from successive images 30 min apart to within approximately 1 K. To facilitate the interpretation of the clear sky brightness temperatures, a simplified model of the radiative transfer at 6.7-mum is presented. This model, based upon a set of irregularly spaced, strongly absorbing, pressure-broadened lines, demonstrates that accurate to within approximately 1 K or 10% of the actual relative humidity, the brightness temperature at 6.7-mum is proportional to the natural logarithm of the appropriate vertical average of the relative humidity divided by the cosine of the viewing zenith angle. Estimates of upper tropospheric water vapor inferred in this way from GOES E observations are presented for July 1987. The geographic distribution reflects many well-known features of the large-scale atmospheric circulation. A clear dependence of the greenhouse effect of upper tropospheric water vapor upon the large-scale dynamics is also demonstrated. Finally, the observed relationship between the upper tropospheric relative humidity and the occurrence of upper tropospheric cloud cover is presented and its implications for the parameterization of clouds in general circulation models are discussed.