Using an adapted version of the MIRAGE model, this paper aims at assessing the impact of a widespread liberalization in agriculture, as proposed in the revised Harbinson proposal. The CGE model includes imperfect competition and increasing returns to scale in industry and services. It assumes land and labor mobility to be imperfect across sectors, and developing countries have a dual labor market. Special emphasis is put on measuring properly protection and domestic support. Domestic support data is updated to 2001 for the EU and the US, and accounts for the Agenda 2000 reform and the New Farm Bill. Protection data, from the MAcMap-HS6 database, describes bound and applied tariffs, taking preferential agreements exhaustively into account. The liberalization hypotheses used in each scenario are applied at the HS-6 level. The results provide a downward appreciation of multilateral agricultural liberalization and a contrasted picture of the benefits developing countries may draw from this process.