Micro-simulation is an approach to analyze the impact of economic and social policy on the distribution of target variables, not just on the means. It easily includes the true policy instruments and handles highly nonlinear relations. Most models currently used in policy analysis are static and they do not include behavioral response to policy changes, just their first order effects. There is, however, an increasing demand for dynamic models including behavioral responses. This paper surveys current theory and practice in micro-simulation with an emphasis on behavioral modeling, and discusses issues of model building, data availability, estimation, testing and validation.