Background-Little is known about the resuscitation preferences of patients with heart failure, how they may change over the course of the disease, and their association with mortality. Methods and Results-We enrolled consecutive Southeastern Minnesota residents with heart failure from October 2007 through September 2011 into a longitudinal study. Information on resuscitation preferences (Full Code or do-not-resuscitate [DNR]) was obtained from medical records through April 1, 2013. Of 608 patients enrolled, 237 died during follow-up. At enrollment, most patients (73.4%) were Full Code, whereas at death, most (78.5%) were DNR. The independent predictors of DNR status at enrollment were advanced age, chronic obstructive pulmonary disease, previous malignancy, and decreased mobility. Patients who were DNR were at increased risk of death (unadjusted hazard ratio, 2.03; 95% confidence interval, 1.48-2.73; P<0.001), but this risk did not persist after adjusting for age, comorbidity, and self- perceived general health (hazard ratio, 0.97; 95% confidence interval, 0.74-1.30; P=0.83). Of 481 patients who were Full Code during follow-up, 22 (4.6%) received cardiopulmonary resuscitation for an in-hospital cardiac arrest. Eight patients survived to hospital discharge; only 2 (9.1% of those receiving cardiopulmonary resuscitation) made a complete recovery and returned home. The median time from a final decision to be DNR until death was only 37 (7,70) days. Conclusions-The resuscitation preferences of patients with heart failure seem to be driven by the decline in clinical status that often accompanies advanced age and multimorbidity. Furthermore, these data suggest that electing DNR status does not independently affect a patient's risk of death. © 2014 American Heart Association.