The effects of the customs union are ambiguous and depend on different factors. The theoretical ambiguities raise a need for empirical studies to generate information for a particular case. This study looks on the effect of EAC-CET for selected agro-food sensitive products of Burundi’s trade, welfare and tariff revenue using the partial equilibrium model; specifically, the SMART model at HS-6 level. The analysis makes use of trade data of 2010-2016 from WITS database, and the EAC-CET schedule was obtained from EAC-CET document version 2017. The study defined 2 tariff scenarios. The first is a CET on selected sensitive products imported by Burundi from the rest of the world, and the second scenario assessed the variation of CET in tariff revenue. Results indicate that the implementation of EAC-CET led to a decrease of imports from the rest of the world, which created losses in a trade estimated to be 6 124 and 3 3782 (thousands US$) in rice and wheat, respectively. This also led to a diversion of imports of rice and wheat to its EAC partners estimated to be 1 626 and 831 (thousands US$), respectively. Further results indicate that gains of tariff revenue for the Burundian government due to high tariff on rice and wheat, are respectively estimated to be 9 277 and 6 627 (thousands US$), but Burundi would gain extra 231 and 363 (thousands US$) and these losses are due to the variation of CET. In terms of welfare, Burundi loses in terms of consumption of rice and wheat 1 258 and 6 051 (thousands US$) but gains in maize. The study recommends removal of rice and wheat among sensitive products and suggests that the addition and removal of products in the list of sensitive should be based on their welfare implications and needs of local consumers.