Few studies have examined the outcome of do-not-resuscitate (DNR) patients in surgical intensive care units (SICUs). This study deliberated the association between a DNR decision and the increased risk of death methodologically and ethically. This study was conducted in three SICUs. We collected patients' demographic characteristics, clinical characteristics, and the status of death/survival at SICU and hospital discharge. We used Kaplan-Meier survival curves to compare the time from SICU admission to the end of SICU stay for the DNR and non-DNR patients. Differences in the Kaplan-Meier curves were tested using log-rank tests. We also conducted a Cox proportional hazards model to account for the effect of a DNR decision on mortality. We found that having a DNR order was associated with an increased risk of death during the SICU stay (aRR = 2.39, p < 0.01) after adjusting for severity of illness upon SICU admission and other confounding variables. To make the conclusion that a DNR order is causally related to an increased risk of death, or that a DNR order increases the risk of death is absolutely questionable. By clarifying this key point, we expect that the discussion of DNR between healthcare professionals and patients/surrogate decision-makers will not be hampered or delayed.