This article develops an index of trade barrier for 119 countries by using the tariff and the import data from TRAINS database of UNCTAD at 6-digit level of HS classifications. Bivariate as well as multivariate econometric models have been estimated to explain cross-country variations in the constructed trade barrier indices. The results show that cross-country variations in trade barrier indices are much influenced by variations in per capita income, population and literacy rate. This article considers Bangladesh as an example and finds that at disaggregated levels of commodity classification, there are significant differences among trade barrier indices across commodities and cross-country rankings of trade barrier indices are not necessarily consistent with the rankings of trade barrier indices in the cross-industry context.