One of the most important steps of surface-related multiple elimination (SRME) is subtraction of the matching multiple waves from the original data. The L2-norm adaptive multiple subtraction method applies to the situation in which the energy of primary wave is very low; however, the L1-norm adaptive multiple subtraction method could apply to the situation in which the energy of primary wave is strong. To obtain a more ideal method, the authors built a mathematical model to combine L2-norm and L1-norm adaptive multiple subtraction method by means of weighted array, and proposed the method for determination of weight coefficient. In addition, the method which combines synthetic data with field data was illustrated, and the results between using the hybrid L1/L2-norm and using L2-norm and L1-norm alone were comparatively studied. The results show that using the hybridL1 /L2-norm adaptive multiple subtraction method can not only suppress the multiple wave better, but also keep the primary wave well.