Game-Theoretic and Rational-Decision models are widely adopted not only for theoretic purposes, but also in order to inform policy advice. Experimental techniques to perform controlled tests of such models have flourished during the last fifteen years. Nonetheless, we lack almost completely an investigation in the methodology of experimental economics. I intend to give a detailed account of the current work in modelling economic situations, and of its methodological foundations and presuppositions. The main views about modelling in the social sciences will be assessed, and several representative case studies will be developed in order to draw some conclusions about the best way forward for empirical economics. Particular attention will be devoted to the peculiarities of modelling in the social as opposed to the natural sciences, and of building experimental models of behaviour as opposed to highly theoretic models.