When forecasting the impact of novel policy interventions, simulations are standard. If the behavior of the entire system is complex and not well identified by existing data, simulations that focus on the behavior of smaller units, such as individuals, may be preferred. So-called microsimulation models can incorporate complications such as clustering, nonlinearity, non-standard distributions, and time-dependence. This talk will present an overview of microsimulation techniques, with a focus on statistical features and dynamic (vs static) simulation, especially in health policy settings. I will also describe the current development of a model of health insurance coverage and health care spending of Medicare beneficiaries.