Abstract: In the face of mainlobe jamming， current anti-jamming methods declines sharply， in order to solve this problem， a space-polarization domain combined anti-mainlobe jamming method based on sparse reconstruction is proposed in this paper. Firstly， sparse reconstruction is used to estimate the direction of arrival （DOA） of the interference signal. Then the polarization parameters of the interference signal are estimated by using the principle of maximum correlation and the constraint condition of constant mode. Finally， a joint oblique projection filter is constructed to suppress interference by using the estimated signal steering vector. Simulation experiments show that this method can effectively suppression mainlobe interference signals and sidelobe signals. And this method has a higher output Signal to Noise Ratio （SNR） than the current popular anti-jamming methods， especially when it is used to suppress jamming signals that are very close to the desired signal in the airspace. In addition， the new method has a high computing speed and meets the practical requirements of radar anti-jamming.