The image quality and image interpretation accuracy of Synthetic Aperture Radar (SAR) deteriorates due to the pre-sence of Radio Frequency Interference (RFI) in space. Based on the weighted least squares iterative adaptive algorithm, a non-parametric iterative adaptive RFI suppression is proposed. The proposed approach supplements a pre-detection step for RFI before imaging processing. After the initial detection of data with RFI, only the data with RFI is subjected to SG (Savitsky-Golay) convolutional smoothing to improve the detection ability of weak RFI. After filtering out the interfering bands by using a band-reject filter, iterative adaptive processing is employed to estimate the signal components of the removed bands to obtain the desired spectrum in the range direction. Compared with the traditional filtering approach for RFI suppression in range frequency domain, the proposed approach is more effective in image detail preservation, avoiding the increase of side lobes caused by spectral discontinuity, and has higher computational efficiency.