We develop a general equilibrium model for evaluating domestically financed transfer programmes and derive analytical expressions which provide a framework for combining results from a computable general equilibrium model with disaggregated household data. We separate the welfare impact into three components, i.e. "redistribution", "reallocative", and "distortionary" effects. We show how these are subsumed within one parameter, the "cost of public funds". Using a Mexican programme for illustration, we show that substantial welfare gains result from a switch from universal food subsidies to targeted transfers, reflecting both the improved targeting efficiency of the latter and the relaxation the trade-off between equity and efficiency when designing tax systems. Copyright 2004 Royal Economic Society.