The ability of operational radiative transfer models to accurately predict remote sensing instrument observations (e.g., calibrated radiances) over a wide variety of geophysical situations is critical to the performance of trace gas retrieval algorithms. As part of the validation of the Measurements of Pollution in the Troposphere (MOPITT) satellite instrument, we present a technique for comparing operational calibrated thermal band (4.7 ¿m) Earth-view MOPITT radiances with corresponding values calculated using the MOPITT operational radiative transfer model. In situ carbon monoxide (CO) profiles sampled from aircraft in coordination with MOPITT overpasses serve as the foundation for MOPITT validation. Characteristics of radiance errors due to in situ sampling characteristics, CO temporal and spatial variability, and surface emissivity are discussed. Results indicate that radiance biases for most of the MOPITT thermal channel radiances are typically on the order of 1%. Observed radiance biases are largest and most variable for the pressure modulation cell difference-signal radiances, probably because of the lack of in situ data in the upper troposphere and lower stratosphere.