Despite limited data on patient outcomes, simulation training has already been adopted and embraced by a large number of medical schools. Yet widespread acceptance of simulation should not relieve us of the duty to demonstrate if, and under which circumstances, training learners on simulation benefits real patients. Here we review the data on performance of healthcare providers or trainees following simulation training, and discuss ways of enhancing transfer of learning from simulated to real patients. While there is tremendous potential for simulation in medical education and healthcare, further studies are needed to identify if and when simulation training improves the quality of care delivered to patients, and to compare the cost-effectiveness of simulated learning experiences to lower fidelity and less expensive interventions.