A computational solution to the inverse sinusoidal input describing function (SIDF) for a broad class of static nonlinearities is presented. The proposed numerical solution uses the SIDF gain and phase distortions to identify the nonlinearity. This solution, unlike existing methods in the literature, does not require a priori knowledge of the nonlinearity structure in the estimation process and is applicable to both single- and double-valued nonlinearities. The output from the algorithm is a nonparametric model of the nonlinearity from which a parametric model can be recovered by least-square estimation (LSE) method. Three examples are presented to validate the proposed algorithm.