In the low signal-to-noise ratio seismic data,linear interference suppression is one of the key processing chains.The traditional method of linear interference suppression in frequency domain tends to generate the phenomena of aliasing and earthworms.Therefore,the authors propose a filtering method based on Cadzow to suppress linear interference.Firstly,the seismic data will be transformed by linear function in the time-space domain.Secondly,according to the Cadzow filtering method,a two-dimensional Hankel matrix is created,and then singular value decomposition is run.Finally,by means of reducing the rank of matrix, linear interference is suppressed.The theoretical model and field data experiments show that the method can effectively remove linear noise and protect the effective wave,thus significantly improving the signal-to-noise ratio of pre-stack seismic data.