This article analyses the trade impact of preferential trade agreements (PTA).  We firstly revisit the literature using a “traditional gravity” setting extended to rely on detailed data at the sector level (26 ISIC industries), using a panel of more than 100 countries between 1976-2000.  Secondly we use the border effect methodology. Lastly, dyadic fixed effects make it possible to control for unobserved characteristics of country pairs, hence for the endogeneity of the PTA.  We systematically disentangle the various arrangements and tentatively introduce tariffs.  The positive trade impacts of the EU, NAFTA and ASEAN are downsized by such improvements.