Abstract Measured fault/slip data are commonly heterogeneous in nature. A critical issue of stress inversion is how to separate them into homogeneous subsets. In the parameter space, data vectors in each subset tend to distribute toward a hyperplane, whose solution of stress vector we solve for. The separation procedure therefore becomes recognition of linear structures of the entire data set. In this paper, a MS-Fortran code is presented for stress estimation using the fuzzy C-lines analysis technique. It is automatic in looking for the optimal stress estimates that have the largest partition coefficients within a specified range of division number. It is finally applied to one example of the feasibility of the inversion method.