We learn in medical school to avoid quest ions from patients and families requiring that we hypothesize what we would do in their situation, if our own health or that of a family member hung in the balance. It seems na tu ra l to avoid such speculation. There is something undeniab ly unset t l ing about insert ing oneself as the proxy decision maker. A little more t han a year ago, during the final year of my residency, I completed a 2-week assignment as the night-float senior resident at the Veterans' Hospital affiliated with our university. Such night-fioat shifts were developed to comply with 80-hour work-week restrictions. While not the busies t of our educational rotations, it was perhaps the one in which we shouldered the greatest clinical responsibility. As the senior Internist overnight supervising 2 interns, I was on my own, without the reassur ing proximity of fellows and coresidents at the university hospital. During my last night of the 2-week assignment , several pat ients became acutely ill s imultaneously. I was on call with 2 very conscientious interns, who were still only 1 month removed from medical school. Every quest ion they posed to me that night had to do with a seriously ill pat ient whom I felt uncomfortable not assessing myself. One of these was a 38-year-old patient, J.L., who was 150 days removed from a bone marrow t ransp lan t for relapsed acute leukemia. He was admitted the previous night for disseminated skin lesions, felt to be secondary to an as yet undiagnosed infection. He was being treated empirically with broad-spectrum antibiotics and had undergone multiple skin biopsies to clarify his diagnosis. When I arrived for my shift, the on-call senior resident I was replacing told me that J.L.'s systolic blood pressures were in the 70s, and that the in tern was at tending to this. Soon after, I was called away to a hypoxic pat ient requiring intubat ion, bu t then was called again by the intern, who reported that J.L. 's blood pressure had not improved. I went to see him. He was alert and oriented, bu t I felt that he belonged in the ICU. However, because no ICU beds were available, we moved him to a telemetry bed for closer monitoring. We planned to aggressively replete his int ravascular volume, and re-evaluate. I was then called away to other calamities and heard nothing more, which I took to mean our plan was successful. When I finally had a chance to see J.L. again early in the morning, his blood pressure had not improved. I told him that we needed to transfer him to the ICU. He declined. He was sitting up, speaking coherently, and answering quest ions appropriately. I updated the in tern and at tending from the primary service. The at tending had a long relationship with J.L. dat ing to before his t ransplant , and knew he and his wife had always wanted everything possible to be done for him. He regarded J.L.'s refusal of transfer to the ICU as evidence he was confused. Indeed, despite his seeming lucidity, J.L. did not recognize him. We arranged his transfer in short order. Many days later, he died in our ICU. Despite the intervening year and a half, I cont inue to have second thoughts about J.L. and the adequacy of the care I provided for him that night. What seemed to be individually reasonable decisions amounted to the wrong decision in sum. It is hard to not feel guilty when a pat ient remains hypotensive all night on one's watch. Residents often treat shift work differently than their responsibilities as the primary team. It is easy to be lulled into a "keep them alive unti l morning" mindset, and hard to s u m m o n the discipline to quest ion decisions made by others during the day. J.L. was seen by 2 senior residents before my arrival, making it easy for me to feel that his hypotension had been "dealt with," al though clearly it had not. I wish we had transferred J.L. to the ICU when he first became hypotensive, even if it mean t transfer to the university hospital. I wish I had communicated more clearly to the in tern caring for him to notify me when his blood pressure did not improve. I wish I had been quicker to recognize his confusion when faced with his seemingly lucid refusal of transfer to the ICU.