In this paper, I extend Ossa (2011) to study the impact of the Most-Favored-Nation principle (MFN) of GATT/WTO on the characterization of Pareto-improving bilateral trade agreements. The paper offers four main model predictions. First, I characterize bilateral trade agreements with third-country tariff adjustments that improve the welfare of negotiating countries while preserving the welfare of the outside country unchanged. Second, MFN guarantees that a bilateral trade agreement always improves the welfare of outside country and potentially causes a free- rider problem. Third, MFN can be too restrictive and prevent possible Pareto-improving trade agreements if initial tariffs are generally low. Lastly, this paper suggests a reason of why free trade agreements (FTAs) are more desirable than bilateral trade agreements under MFN, when current tariffs are sufficiently low. I quantify the firm-delocation effects and welfare changes in three counterfactual situations: (i) a bilateral trade agreement without third-country tariff adjustments, (ii) a bilateral trade agreement under MFN, and (iii) moving to a global free-trade economy. The quantitative results support the model predictions.