The effect of do-not-resuscitate (DNR) orders on physicians' decisions to provide life-prolonging treatments other than cardiopulmonary resuscitation (CPR) for patients near the end of life was explored using a cross-sectional mailed survey. Each survey presented three patient scenarios followed by 10 treatment decisions. Participants were residents and attending physicians who were randomly assigned surveys in which all patient scenarios included or did not include a DNR order. Response to three case scenarios when a DNR order was present or absent were measured. Response from 241 of 463 physicians (52%) was received. Physicians agreed or strongly agreed to initiate fewer interventions when a DNR order was present versus absent (4.2 vs 5.0 (P =.008) in the first scenario; 6.5 vs 7.1 (P =.004) in the second scenario; and 5.7 vs 6.2 (P =.037) in the third scenario). In all three scenarios, patients with DNR orders were significantly less likely to be transferred to an intensive care unit, to be intubated, or to receive CPR. In some scenarios, the presence of a DNR order was associated with a decreased willingness to draw blood cultures (91% vs 98%, P =.038), central line placement (68% vs 80%, P =.030), or blood transfusion (75% vs 87%, P =.015). The presence of a DNR order may affect physicians' willingness to order a variety of treatments not related to CPR. Patients with DNR orders may choose to forgo other life-prolonging treatments, but physicians should elicit additional information about patients' treatment goals to inform these decisions.