Acoustic seabed classification is significant for the study of the seabed environment and the ecosystem, in which multibeam and sidescan sonars are the most common sonar equipment. However, multibeam and sidescan sonars are both largely affected by radiometric distortion, which makes the noise of the sonar image too big to be eliminated and the seabed sediment misinterpreted. Moreover, the universal seabed classification methods for multibeam and sidescan sonars are still deficient. Therefore, this paper proposes a universal seabed classification method of multibeam and sidescan sonars in consideration of radiation distortion. Steps of the method include correction of radiometric distortion, normalization of both mulitbeam and sidescan sonars data, adaptive unsupervised seabed classification, and morphological image denoising. A complete universal seafloor classification procedure was also presented. The proposed method was applied to the same water area in Fujian Coast, where multibeam and sidescan sonars data were both measured and the seabed classification results of the water area were obtained. Experimental results show that the seabed classification results processed from multibeam and sidescan data in the same water area were in high consistency, which verifies the validity of the universal seabed classification method and improves the reliability of the classification results through mutual verification.