Terahertz is expected to be the best waveband for remote sensing of cirrus. However, the retrieval results of satellite terahertz passive remote sensing are susceptible to different atmospheric conditions. The terahertz radiation was simulated under the clear sky and the cloudy situations, based on the atmospheric radiative transfer simulator. The influences of the surface reflectivity, atmospheric profiles and the middle-low level water clouds to terahertz radiation were analyzed. The retrieval errors caused by these factors were calculated based on the multiple lookup-tables method. The results indicate that the influence caused by changing the surface reflectivity mainly range below 300 GHz, and there is no influence on the selected channels. Among the atmospheric profiles, only changing water vapor profiles and temperature profiles have effects on the retrieval results. When the temperature profile changes are between ±2 K, or the water vapor profile changes are between ±20%, for cirrus whose particle size is above 50 μm and ice water path is above 10 g/m2, the retrieval errors stay below ±20%. For low-level water clouds (cloud base height below 2 km), the effects on retrieval errors can be ignored. The retrieval errors caused by middle-level water clouds increase with the increases of the cloud base height and cloud optical depth.