The purpose of this study is two fold. First, to estimate the impact of institutional and non-institutional arrangements on bilateral trade, and second to analyse the impact of SAFTA on bilateral trade in the short as well as in the long run. The empirical analysis which is based on the panel of eight South Asian countries, comprising data over the period i.e. 1975–2013 is conducted using fixed effects model along with Pooled Mean-Group (PMG) estimator for estimating the short and long-run relationships. The analysis has shown that trade agreements including South Asian Free Trade Area (SAFTA) and the Most Favoured Nation (MFN) are not effective in promoting trade, due to low institutional quality and stringent non-institutional arrangements, including high tariff along with low physical infrastructure. Further empirical analysis has shown that both SAFTA and MFN can only contribute to bilateral trade significantly, if complemented by institutional framework. As a policy lesson, to improve the trade ties between India and Pakistan, improvement in physical as well as soft infrastructure is required. Any trade agreements between the two, including MFN can only be effective, when it is supported by a well-defined and enforced institutional framework that ensure the implementation of policy reforms needed to reduce tariff rate and remove non-tariff barriers.