Computable General Equilibrium (CGE) models are often used to simulate the effects of political reforms. In these models the way demand reacts to price and income changes is of crucial importance when one wants to evaluate the evolution of demand in the baseline or simulate the effects of political reforms. Yet, functional forms used to model households’ demand in CGE models do not necessarily exhibit enough flexibility to fully acc ount for income and price changes on the structure of demand. Our objective in this study is to empirical ly compare the results generated by a CGE using different demand functions, to see if the additiona l complexity associated to the flexibility of demand really modifies the results and has an impact on po licy recommendations. We implement four demand systems in the Mirage model: a Linear Expenditure-C onstant Elasticity of Substitution (LES-CES) function, a Cobb-Douglas function, a Constant Elasticity of S ubstitution (CES) function, and a Normalized Quadratic Expenditure System (NQES) demand system. We conduct simulations of trade reform with a particular focus. We then compare the economic effects of the reform simulated with the different demand functions. From these first results, the LES-CES thus appears to be a good compromise between flexibility and simplicity. It actually requires less parameters and simplifies the model solving compared to the NQES to get similar results.