Objective: To assess whether cardiopulmonary resuscitation performed by in-house physicians is effective for long-term-care residents. Design: Retrospective chart review. Setting: Long-term-care facility with an intermediate care unit, "skilled" care unit, and a convalescent and assessment unit at a retirement community for veterans. Participants: All residents resuscitated from April 1987 to August 1990. All participants were male. The mean age was 75 years +/- 7.3 (range 42-93 years). Main Outcome Measurements: Charts were abstracted for demographics, advanced directives information, information about the arrest, and post-resuscitation course. Results: Forty-five elderly residents underwent resuscitation during this period. Nine residents (20%) were successfully resuscitated, with seven dying within 24 hours of hospitalization. No residents survived to return to long-term care (95% CI, 0-7%). The diagnoses were consistent with age-related chronic disease. Seventeen (38%) arrests were witnessed. The predominant rhythm at onset of resuscitation was asystole. Conclusion: We conclude that cardiopulmonary resuscitation, even when performed by a trained and experienced physician and team, has limited benefit for elderly long-term-care populations.