Subject and Purpose. The writers aim at developing and testing a new method of interference mitigation, proceeding from the example of radio emissions from Jupiter. Its effectiveness is compared with the results of other workers, equally relating to the case of a significant overlapping, within the time-frequency window under analysis, of the areas occupied by the useful signal and the interference. Methods and Methodology. The analysis has revealed several fundamental limitations associated with the use of standard statis- tical methods for identifying sources of interference. A new approach is proposed that allows separating useful signals from inter- ference in the time-frequency plane. It is based on the idea of transferring the statistical analysis from the space of signal amplitudes to such of linear patterns which are formed by maximal readings while the spectrograms are being scanned in time, frequency or otherwise. Results. Methods of statistical data processing have been suggested which allow analysis of signals of a variety of power lev- els against the background of interference of comparable intensity. This enables a detailed analysis of the time-frequency patterns demonstrated by signals with a broad range of parameter variations. The algorithms developed demonstrate stability against changes in the interference background conditions that may be caused either by human activity or by natural factors, such as, e.g. ionospheric perturbations, changes in the frontend frequency response of a receiver resulting from a changed antenna beam orientation, or else from Faraday’s polarization plane rotation in the radio emission being received. Conclusions. The necessity of creating new interference mitigation techniques is stipulated both by worsening of the general level of interference at radio frequencies, and by the growth of complexity in the sporadically emerging time-frequency patterns that result from the improved time and frequency resolutions in the course of signal reception. A significant progress has been achieved, owing exclusively to the fundamental modifications of the signal processing algorithms that are based on varying the direction of analysis in the time-frequency plane.