Cardiopulmonary resuscitation (CPR) is often performed in modern critical care units, but its efficacy has not been evaluated in this setting. It is important to evaluate CPR in critical care units because these patients often have multisystem disorders and suffer from diseases reported to carry a poor outcome after CPR. Inappropriate resuscitation of patients in this setting results in increased cost of care (both financial and emotional), with little tangible benefit. To address the question of successful resuscitation in the medical intensive care unit (MICU), we retrospectively reviewed the records of 114 patients who underwent CPR in our MICU over a three-year period. Eighty patients (70 percent) were not successfully resuscitated, 21 patients (18 percent) were successfully resuscitated but died before discharge, and 13 patients (11 percent) survived to leave the hospital. We evaluated a number of prearrest conditions (diagnoses, age, sex, duration of hospitalization, length of ICU stay, and severity of illness as measured by APACHE 2 scores) and arrest conditions (the initial cardiac rhythm and duration of CPR) to determine if the outcome after CPR was influenced by any of these parameters. Among the prearrest conditions, only a diagnosis of hypotension or sepsis and an elevated APACHE 2 acute physiology score were independently associated with a poor outcome after CPR. The only arrest condition found to be independently associated with outcome following CPR was the duration of resuscitative effort (p less than 0.01). The patients who were successfully resuscitated but died before discharge were not different from the patients who were not successfully resuscitated in any parameter that we evaluated. These results demonstrate that CPR can be successful in the MICU and that there are prearrest and arrest parameters which are useful in identifying those patients most likely to benefit from CPR in the critical care setting.