Representing the radiative impact of water vapor accurately, especially in the upper troposphere, is an essential component of effective global climate simulations. The direct comparison of modeled and observed spectral radiances provides a detailed means of assessing the simulation of radiative processes in a global climate model with satellite measurements. To evaluate the simulation of water vapor in the National Center for Atmospheric Research (NCAR) Community Atmosphere Model (CAM3) by comparing modeled and observed spectral radiances, CAM3 has been modified to include the Optimal Spectral Sampling (OSS) algorithm developed at AER. OSS has been prepared to calculate radiance spectra for all 2378 channels of the NASA Atmospheric Infrared Sounder (AIRS) spectrometer, which has been measuring infrared radiance spectra from space since 2002. Clear sky AIRS radiances are simulated with OSS for January and July 2004 with two versions of the NCAR CAM3 climate model. These include the original CAM3 and a separate version that uses the RRTMG longwave and shortwave radiation models developed at AER. The latter configuration also applies the Monte-Carlo Independent Column Approximation (McICA), which is an efficient, statistical technique for representing sub-grid cloud variability. Over various geographic regions, selected clear-sky modeled spectra from each simulation are compared to observed, cloud-cleared AIRS radiance spectra. Within the spectral region of relevance to middle and upper tropospheric temperature (700-750 cm-1), differences between modeled and observed brightness temperatures are generally less than 2 K and are mostly insensitive to the radiative transfer method. However, within the spectral region dominated by water vapor absorption (1340-1580 cm-1), brightness temperature differences of up to 5-10 K are present that vary spectrally and by geographic region. Differences in brightness temperature of this magnitude represent significant discrepancies in upper tropospheric water vapor amount of 50% or more in some regions in CAM3. While the brightness temperature differences in the water vapor spectral band are somewhat sensitive to the improved radiative transfer in some geographic regions, it is apparent that the treatment of radiation is not the primary cause of the simulated water vapor discrepancies.