Abstract: We propose a coseismic slip distribution inversion method based on variance component estimation (VCE), this method not only determines the weight of different data sets but also can get the smoothing factor value. To illustrate and validate this inversion method, we conducted inversions on synthetic data sets. Based on the Bayesian framework and according to the posterior probability density distribution of the unknown parameters, the proposed solution is comparted to one of the linear-non-linear inversion solutions to further demonstrate the use of the proposed method. We conduct inversions on the 2003 Bam earthquake with the proposed method and linear-non-linear inversion method, the spatial cross correlation of the resulting slip distribution was about 0.999, the depth of the max slip of the both were 3.04 m and the depth had little difference, the former was 4.50 km, the latter was 4.47 km. The mean slip in the former was 0.714 m, the latter was 0.718 m. But the inversion time of the former was 227 s, and only about 1/8 000 of the later (reaching 1.5×10⁶ s, about 17 d).These results show that the VCE method is effective and faster than the linear-non-linear inversion method.