Rapid techniques for self-potential (SP) data interpretation are of prime importance in engineering and exploration geophysics. Parameters (e.g. depth, width) estimation of the ore bodies has also been of paramount concern in mineral prospecting. In many cases, it is useful to assume that the SP anomaly is due to an ore body of simple geometric shape and to use the data to determine its parameters. In light of this, we describe a rapid approach to determine the depth and horizontal width of a two-dimensional plate from the SP anomaly. The rationale behind the scheme proposed in this paper is that, unlike the two- (2D) and three-dimensional (3D) SP rigorous source current inversions, it does not demand a priori information about the subsurface resistivity distribution nor high computational resources. We apply the second-order moving average operator on the SP anomaly to remove the unwanted (regional) effect, represented by up to a third-order polynomial, using filters of successive window lengths. By defining a function F at a fixed window length (s) in terms of the filtered anomaly computed at two points symmetrically distributed about the origin point of the causative body, the depth (z) corresponding to each half-width (w) is estimated by solving a nonlinear equation in the form ξ(s, w, z) = 0. The estimated depths are then plotted against their corresponding half-widths on a graph representing a continuous curve for this window length. This procedure is then repeated for each available window length. The depth and half-width solution of the buried structure is read at the common intersection of these various curves. The improvement of this method over the published first-order moving average technique for SP data is demonstrated on a synthetic data set. It is then verified on noisy synthetic data, complicated structures and successfully applied to three field examples for mineral exploration and we have found that the estimated depth is in good agreement with the known value reported in the literature.