PURPOSE: To determine the feasibility of high-fidelity simulation for studying physician decision making for critically ill elders. Specifically, we assessed physicians' perceptions of realism and documented their treatment decisions. METHODS: Thirteen hospitalists, 8 intensivists, and 6 emergency physicians role-played the evaluation and management of a case in which the patient and wife were acted by standardized patients (SPs), the vital signs tracings revealed progressively life-threatening hypoxia, and the medical chart data suggested cancer progression as the probable clinical explanation. The SPs' stable, underlying preferences were to avoid ICU admission and intubation; their principal concern was the patient's dyspnea. SPs were trained to respond to questions rather than volunteer information, to mirror physician statements, to defer to unilateral physician recommendations, and, if provided with a choice among treatments, to choose the least intensive option. Physicians' final treatment plans were assessed by 2 independent raters. After the simulation, each physician completed a self-administered survey to assess demographics, case perceptions, risk attitude, reactions to uncertainty, need for justification, and regret. Each physician then underwent a debriefing exit interview. RESULTS: Subjects were experienced physicians, with a mean of 15.3 years since medical school graduation (range 4-41). All (100%) reported that the case and simulation were highly realistic, and their diagnostic and prognostic assessments were consistent with our intent. Eight (29.6%) physicians admitted the patient to the ICU. Among the 8 physicians who admitted the patient to the ICU, 3 (37%) initiated palliative care, 2 (25%) documented the patient's code status (DNI/DNR), and 1 intubated the patient. Among the 19 physicians who did not admit the patient to the ICU, 13 (68%) initiated palliative care and 5 (42%) documented code status. In a logistic regression model using hospitalists as the referent category, intensivists (OR=12, p=0.048) and emergency physicians (OR=12, p=0.060) were more likely to admit the patient to the ICU. Years since medical school graduation were inversely associated with the initiation of palliative care (p=0.043). There were no predictors of code status documentation. CONCLUSIONS: Hospital-based physicians from the same institution faced with an identical simulation scenario demonstrated significant variation in treatment decisions. Simulation is a feasible methodological approach to studying physician decision making.