Although approximately one of five people in the United States die in nursing homes (NHs), little has been written about their quality of dying, including the quality of terminal medical care. The purpose of this study is to review actual medical practices in NHs to suggest factors important for delivering good quality terminal care. Four NHs were surveyed for management of residents who died in 1992. A convenience sample of charts of newly admitted and longer term residents were abstracted for demographic variables, death, diagnostic categories, and various laboratory and other parameters. Charts of those residents who died were further reviewed using indicators of quality medical care, such as presence of advance directives, control of pain, and control of dyspnea, based upon recent published clinical practice guidelines for terminal care in NHs. Three hundred and seventy-one charts were abstracted. Forty-one charts documented the resident's death. We found that NHs without regulatory difficulties usually had expected deaths that were managed approximately as measured by terminal medical care quality indicators. NHs with a history of regulatory difficulties had a higher prevalence of residents who died suddenly and unexpectedly, often with problems in the quality of care as measured by the same indicators. There was a correspondence between physician certification, antemortem diagnosis of terminal illness, and appropriate terminal care. We conclude that physicians are able to recognize impending death and redirect the medical care of dying NH residents toward goals of terminal care management. This is more likely to occur in a NH environment that places greater emphasis upon total quality management. We suggest that another indicator in providing good NH terminal care is the physician's performance in predicting a short life expectancy.