Available estimates of biofuel-induced land use change (LUC) and corresponding greenhouse gas (GHG) emissions vary on a wide range while estimations obtained from each model are highly sensitive to certain assumptions and key parameter values. Available studies often suggest that biodiesel and ethanol and/or EU and US biofuels would lead to different LUC and GHG emissions but they don’t agree on the type and/or the origin of the biofuel which would induce the least LUC and GHG emissions. In this paper we investigate the reasons behind this feature. We show that the Armington modeling of trade flows, which is currently used in models, contributes to this pattern. Using both observed data and the partial equilibrium model MATSIM-LUCA, we show that LUC stemming from the development of biofuels is highly dependent on assumptions made on trade: the Integrated World Market (IWM) approach, which relies on the homogenous product assumption, tends to erase differences in estimates of induced LUC from biodiesel and ethanol and from EU and US biofuels as compared to the Armington approach, that postulates that product are differentiated according to their origin and thus less substitutable.