Matrix element reweighting is a powerful experimental technique widely employed to maximize the amount of information that can be extracted from a collider data set. Given a set of theoretical hypotheses and a sample of experimental events, the method assigns a weight to each hypothesis on an event-by-event basis and in this way provides a discriminator between different theoretical assumptions. The computation of the weights is intricate, because it involves a difficult convolution of the theoretical information on the hard scattering with the experimentally available information on the final state. In this talk, I will present a general algorithm aimed at evaluating the weights appearing in the matrix element method for any process of interest in the standard model and beyond. The implementation builds on MadGraph, and is completely automated. I will also discuss a few sample applications that show the capabilities of the code and illustrate the possibilities for new studies that such an approach opens up. (The talk is based on work presented arxiv:1007.3300.)