The comutation of the equivalent capacitances for three-dimensional (3-D) interconnects features large memory usage and long computing time. In this paper, a matrix sparsifica-tion approach based on multiresolution representation is applied with the method of moments (MoM) to calculate 3-D capacitances of interconects in a layered media. Instead of direct expansion of the charge distribution by the orthogonal wavelet basis functions, the lare full matrix resulting from discretization of the integral equations is taken as a discrete image and sparsified by two-imensional(2-D) multiresolution representations. The inverse of the obtained sparse matrix is efficiently implemented by Schultz's iterative apprach. Several numerical examples are given and the results obtained show that the proposed method significantly sparsify the matrix equation and the capacitance parameters computed by the matrix equation and the capacitance parameters computed by the matrix equation with high sparsity agree well with the results of other reports and those computed by an established capcitance extractor FASTCAP.