This paper reviews the experience of the use of computable or applied general equilibrium (CGE or AGE) models to affect public policy. The range of issues on which CGE models have had an influence is quite wide, and includes structural adjustment policies, international trade, public finance, agriculture, income distribution, and energy and environmental policy. In the cases where CGE models have enlightened the policy debate, the reasons have to do with one or more of the following: (i) consistency between results from CGE models and other types of analysis (for instance in the debate on NAFTA); (ii) the fact that the CGE models captured particular features of the economy, such as some structural rigidities and institutional constraints, that rules of thumb, based on simpler analysis failed to capture; or (iii) CGE models provided a consistent framework to assess the linkages and tradeoffs among different policy packages. We also consider misuses of CGE models in policy debates. Most of these stem from: (i) pushing the model beyond its domain of applicability; (ii) violating the principle of Occam’s razor—use the simplest model suited to the task; (iii) the “black box syndrome”—results whose link with the policy change is opaque. In assessing the use of models in policy, it is important to distinguish between stylized and applied models. Both have been used in policy debates, but there are important differences in their uses, particularly in their domain of applicability. Stylized models tend to be small, narrowly focused, and emphasize a particular causal chain or policy. Applied models are usually larger, seek to capture important institutional characteristics of the economy being modeled, and encompass a wider spectrum of issues; but they are vulnerable to the black-box syndrome and violation of Occam’s razor. Complementary use of stylized and applied CGE models has enhanced the effectiveness of both in policy debates.