A generalized inversion computer program is described for deriving a profile (i.e., a vector of functional values) from a finite set of imperfect measurements. The program assumes the observations are related to the profile by a nonlinear first kind operator equation. The method of solving the nonlinear equations is to iterate the solution of their linear approximation. Linear equations can be solved with this method in one iteration. The program described here uses either the Backus-Gilbert technique or the generalized Landweber inversion technique to solve the linear approximation at each stage of the iteration process.