Linking a global computable general equilibrium model with household surveys of Brazil, Mexico, Colombia and Chile, this paper estimates the initial impacts on the poor of regional and multilateral trade liberalization scenarios. This approach combines the advantages of using general equilibrium consistent changes in factor and good prices with the detailed information on household endowments and preferences provided by the surveys. Furthermore this methodology allows to decompose the total effect on poverty into growth and inequality components. Results show that due to their different initial positions in terms of trade protection, economic structure and poverty levels, the impacts on poverty are quite dissimilar across the four countries studied here, and that is also the case when growth and distributional effects are considered separately. The detailed analysis shows that, even when the aggregate poverty effects are closer, these result from diverse effects of opposite sign that can only be captured by accounting for the full heterogeneity in the household data. Such a richer analysis of the poverty impact may have implications for compensatory policy measures and even for the design of ultimately more successful trade reforms.