During the period of processing and analysis of on-orbit radiometric calibration data received from a space-based infrared camera launched recently, three practical issues were extracted and resolved at the data level, specifically how to exclude the invalid calibration data; how to determine the appropriate on-orbit decontamination time in the presence of increasing contaminants inside the camera system; and how to calibrate images without suitable calibration coefficients. Three major types of invalid data were summarized according to their appearances and possible causes after analyzing data from many on-orbit calibration tests, and the targeted filtering strategies were proposed with proven excellent performance in practice. A two-term exponential model was established to characterize the observed camera degradation by dividing the corresponding digital number into the blackbody and non-blackbody terms. Based on the model, degradation trends of the camera response, radiance resolution, and signal-to-noise ratio were estimated, respectively, to help determine the contamination tolerance from different aspects. Attempts were made to predict calibration coefficients by the pixelwise degradation models, and then they were applied to image calibration. Results show that the predicted coefficients can effectively compensate the calibration errors due to degradation and can be treated as an alternative if time-matched coefficients are unavailable.