183.31 GHz observations from the Advanced Microwave Sounding Unit B (AMSUB) instruments onboard the NOAA 15, 16, and 17 satellites were used to derive a new data set of Upper Tropospheric Humidity (UTH). The data set consist of monthly median and mean data on a 1.5 degrees latitude-longitude grid between 60 degrees S and 60 degrees N, and covers the time period of January 2000 to February 2007. The data from all three instruments are very consistent, with relative difference biases of less than 4% and relative difference standard deviations of 7%. Radiometric contributions by high ice clouds and by the Earth's surface affect the measurements in certain areas. The uncertainty due to clouds is estimated to be up to approximately 10%RH in areas with deep convection. The uncertainty associated with contamination from surface emission can exceed 10%RH in midlatitude winter, where the data therefore should be regarded with caution. Otherwise the surface influence appears negligible. The paper also discusses the UTH median climatology and seasonal cycle, which are found to be broadly consistent with UTH climatologies from other sensors. Finally, the paper presents an initial validation of the new data set against IR satellite data and radiosonde data. The observed biases of up to 9%RH (wet bias relative to HIRS) were found to be broadly consistent with expectations based on earlier studies. The observed standard deviations against all other data sets were below 6%RH. The UTH data are available to the scientific community on http://www.sat.ltu.se. ; 183.31 GHz observations from the Advanced Microwave Sounding Unit B (AMSUB) instruments onboard the NOAA 15, 16, and 17 satellites were used to derive a new data set of Upper Tropospheric Humidity (UTH). The data set consist of monthly median and mean data on a 1.5 degrees latitude-longitude grid between 60 degrees S and 60 degrees N, and covers the time period of January 2000 to February 2007. The data from all three instruments are very consistent, with relative difference biases of less than 4% and relative difference standard deviations of 7%. Radiometric contributions by high ice clouds and by the Earth's surface affect the measurements in certain areas. The uncertainty due to clouds is estimated to be up to approximately 10%RH in areas with deep convection. The uncertainty associated with contamination from surface emission can exceed 10%RH in midlatitude winter, where the data therefore should be regarded with caution. Otherwise the surface influence appears negligible. The paper also discusses the UTH median climatology and seasonal cycle, which are found to be broadly consistent with UTH climatologies from other sensors. Finally, the paper presents an initial validation of the new data set against IR satellite data and radiosonde data. The observed biases of up to 9%RH (wet bias relative to HIRS) were found to be broadly consistent with expectations based on earlier studies. The observed standard deviations against all other data sets were below 6%RH. The UTH data are available to the scientific community on http://www.sat.ltu.se .