Abstract: The next-generation geostationary satellite FY-4A equipped with Advanced Geostationary Radiation Imager (AGRI) was launched on 11 December 2016. In this study, the biases of seven infrared channels (08-14 channel) of AGRI from model simulations are characterized by using the Radiative Transfer for the TIROS Operational Vertical Sounder (RTTOV). The Moderate-Resolution Imaging Spectroradiometer (MODIS) cloud mask is used for selecting clear-sky data. The significant systematic deviations at channel 10 (7.1 μm) are most possibly caused by the overestimation of water vapor in ERA5 (the fifth ECMWF reanalysis) data. The calibration error and higher air temperature of surface layer in ERA5 data are possible causes for deviations at channel 14 (13.5 μm). Since the brightness temperature of near-infrared channel (channel 08, 3.72 μm) is strongly affected by surface reflection, there is a large average bias on the land at channel 08. The bias of this channel over the ocean is less than 0.4 K. The average biases at all other infrared channels over the ocean and land are below 0.6 K and 1.3 K respectively. In addition, surface elevation and sensor zenith also have a certain degree of influence on the simulation and subsequently lead to corresponding bias characteristics. The seasonal variation of bias distribution over the ocean existing at channels 11 (8.5 μm), 12 (10.8 μm) and 13 (12.0 μm) probably comes from the seasonal error of sea surface temperature estimate in the reanalysis data.