Summary Improvement of the signal-to-noise ratio (S/N) of seismic data is necessary in many seismic exploration areas. The attenuation of random noise is an important subject in improving the S/N. Geophysicists usually utilize the difference between signal and random noise in certain attributes, such as frequency, wave number, or correlation. In this paper, we have proposed a novel method utilizing the planar morphological attribute of seismic data to separate signal and random noise. The extraction of the morphological attribute is implemented by the planar morphological operations. The attenuation of random noise is achieved by removing the energy in the smaller morphological scales. We have named our proposed method planar mathematical morphological filtering (PMMF). Application of PMMF on synthetic and field seismic data demonstrates superior performance compared with the 2D median filtering and the singular spectrum analysis (SSA) method.