Two approaches have emerged in the use of CGE models in poverty and income distribution analysis: the traditional representative agent and the microsimulation approach using a large number of households to capture more heterogeneity. This article uses three simple CGE models with representative agent and microsimulation approaches to verify whether the two methods produce compatible results. The use of microsimulation does not appear to lead to different results or contribute significant value-added in impact analysis at the macroeconomic and sectoral levels. However, when poverty and inequality are the focus, the results produced by the two approaches are inverted, revealing that the choice of approach is crucial when the objective of the analysis relates to poverty and inequality.