I arrived on Monday to begin a 2-week stint as attending in our medical ICUs. The first patient was an elderly lady, who had been in hospital for a month with newly diagnosed HIV and Pneumocystis carinii pneumonia. She unexpectedly experienced grand mal seizures and was intubated for airway protection. Brain imaging and lumbar puncture were unrevealing. Her status epilepticus was complicated by aspiration pneumonia-related septic shock, with prolonged hypotension requiring pressors for > 48 h. Acute tubular necrosis developed, which was beginning to resolve by my advent to the service. The patient had been in a deep coma for 3 days since cessation of her seizures. She did not move to deep pain, her pupils were minimally reactive, and she triggered the ventilator occasionally. The house officers presented her as a case of anoxic encephalopathy—the diagnosis suggested by the consulting neurologist. “When did she code?” I queried. “She didn’t” was the response. She remained in her slumber, essentially unchanged, except that she began to move her left arm, nonpurposefully, to deep pain. An EEG demonstrated a paucity of activity—no seizures, but brain waves that portended a very poor prognosis. The consulting neurologist pronounced that there was little or no hope of meaningful survival. Although I felt uncomfortable, not understanding the mechanism of her coma and still suspicious that it was related to multiorgan failure and polypharmacy, the brain waves seemed like incontrovertible evidence that her chances were very poor. I arranged for a family meeting to discuss ongoing management. I have been, historically, the typical “flea”—the last one to give up on patients. And for days, the housestaff and nurses were adamant that we were doing our patient no favors by keeping her alive, languishing on the ventilator.