Is it possible to design a bargaining institution that will substantially improve the international trade negotiations conducted in a multilateral basis? Rausser and Simon [1991] propose a multilateral bargaining (MB) institution which has many attractive theoretical properties, and which appears to be general enough to encompass realistic classes of trade negotiation problems. Harrison, McCabe, Rausser and Simon [1992] provide some elementary "stress tests" of the MB institution in the form of laboratory experiments using contrived preferences. In this paper I extend the "lab stress testing" to assess the outcomes that the MB institution would theoretically achieve using less contrived negotiation problems. In section 2 I present a brief overview of the main properties of the MB institution, including a series of "large" numerical examples. These examples are still contrived in the sense of employing artificial preference structures, but are useful to illustrate the general numerical properties of the model. They provide the backdrop to our simulations using more realistic preference structures appropriate for negotiations over trade policies. In section 3 I review alternative approaches to estimating "realistic" policy preferences. In section 4 I employ one of these methods to generate some preference structures over agricultural trade policy reforms, and solve the resulting MB game.