In this study, a novel 2-stage noise adaptive multilevel fuzzy reasoning–based filter is proposed as a postprocessing filtering technique for noise removal from magnetic resonance images. The edge detection stage uses multilevel fuzzy reasoning to handle uncertainty present in the local information introduced by noise. The image-smoothing stage uses a low-pass filter to retain the low-frequency information of the noisy image. Finally, the output of the edge-detection stage, and the smoothed noisy image are added together to form the noise-free image. The effectiveness of the proposed method in noise removal from MR images is evaluated in terms of visual quality of the image and image quality assessment metrics such as peak signal-to-noise ratio, mean square error, structural similarity index measurement, and image enhancement factor. Experimental results show that the proposed method removes noise more effectively from MR images without destroying finer details and coarser structures as compared with the other tested state-of-the-art methods.