This paper highlights areas of inaccuracy in current modelling techniques for bridge decks with wide cantilevers. A 3-D grilllage model known as the upstand grillage is assessed by comparing its behaviour with that of a three-dimensional, finite-element model. Recommendation are made for the implementation of such a technique. The paper concludes with a comparison of results from analyses of a bridge deck using the standard and upstand grillage analogies and a three-dimensional finite-element method.