Contemporary industries are devoting increasing attention to the product development process, due to tight market shares and the abridged product life cycle. Reliable scaled product testing with rapid prototypes has the potential to improve these processes by replacing traditional costly and time-consuming product tests. In this context, rapid prototypes provide visual, ergonomic, and functional information with minimal time delay. Among the information classes, reliable functional information is least realized because of several features of rapid prototypes: (1) limited material choices and part size; (2) distinct material structure; (3) restrictive loading conditions; and (4) state-dependent material properties. To develop reliable functional tests, an improved similarity method is needed to overcome these limitations. The traditional similarity method, based on a Buckingham П approach, is commonly applied to perform scaled tests. In contrast to this method, wherein the state transformation between two similar systems is derived from dimensional vectors, we present a new similarity method that empirically derives the transformation from a geometrically simple specimen pair. The primary advantage of the new method over the traditional method is the capability to relate highly distorted systems. In this paper, the concept and theoretical framework of the novel similarity method are introduced, and two numerical examples demonstrate the new method.