The study deals with analysis of thermal calibration of the Advanced Very High Resolution Radiometer (AVHRR) aboard National Oceanic and Atmospheric Administration (NOAA) spacecrafts. In particular, we investigate the effects caused by various types of contamination of the thermal calibration data. It is known that the satellite data are prone to corruption. Corruption of the data may occur during transmission through the atmosphere. Data may also be affected or contaminated by some internal processes occurring in the instrument, for example, during digital conversion, fluctuations of the instrument thermal state or solar contamination of calibration cycles. These phenomena lead to perturbations of the true signal that we call unwanted fluctuations. They must be removed or corrected to maximal possible extent. It is shown that methods currently employed in operational practice at NOAA and Canada Centre for Remote Sensing (CCRS) frequently fail to remove some unwanted fluctuations in calibration data that may lead to essential biases in brightness temperatures sometimes exceeding 1K. We propose the method for removing unwanted fluctuations in the thermal calibration data specifically designed for the AVHRR radiometers. The procedure is based on combining the robust statistical procedures and Fourier transform filtering techniques. Application of the method is considered for various components of calibration data: blackbody temperature sensors, blackbody and space count as well as gain and offset in all thermal channels. Local area coverage (LAC) High Resolution Picture Transmission (HRPT) data and Global Area Coverage (GAC) data are analysed. Power spectra analysis of the calibration data has been conducted to estimate impact of various frequency harmonics. Method proposed in the paper maybe useful for the development of calibration techniques for similar radiometers and future NPOESS system.