In this article, a mathematical framework to automatically evaluate the manufacturability of injection-molded and die-cast parts is presented. The framework includes both a logical algorithm for the general problem of feature recognition and an implemented mathematical and numerical algorithm to solve key outstanding challenges in feature recognition for manufacturability analysis. A novel feature recognition method is developed that is based on decomposing the part into elemental cubes and then, making use of their individual manufacturability, the manufacturability of the part as a whole is evaluated. This article discusses a procedure to obtain a 3D binary representation of the solid model in a simple fashion and further develops feature recognition techniques to extract critical manufacturability information from this 3D binary array. The outstanding challenges addressed by the method presented include the finding of parting surfaces, undercuts, holes, and bosses in the context of an injection-molded or diecast part. The algorithm is implemented using a combination of C++ code and Unigraphics solid modeling software. A short example is presented.