This paper suggests spatial models as an alternative to the Armington approach to model bilateral trade. While the use of spatial models has been accepted for decades, they are rarely chosen for such analyses. However, problems inherent in the application of the Armington approach can be overcome through the use of spatial models. To demonstrate, a simple spatial model of the world sugar market is built and used to simulate a multilateral liberalization scenario. Additionally, an identical model is constructed, applying the Armington approach. The results of the spatial model of the sugar market are found to be more plausible than those generated by the Armington-based model.