Decisions to place limitations on the care of patients are complex, and they often involve physicians, other medical professionals, patients, or a surrogate decision-maker, family members, and others. In 1988, the Joint Commission on Accreditation of Health Care Organizations (JCAHO) and the New York State government adopted two different approaches to this complex issue of do-not-resuscitate (DNR) orders: one involved professional self-regulation, whereas the other mandated a standardized procedure requiring completion of legal documents. This study examines the impact of these two different approaches to writing of DNR orders for adult intensive care unit (ICU) patients on utilization and resulting length of stay. The study used three data bases. One is from a larger study designed to update the Mortality Probability Model (MPM), a measure of severity of illness for ICU patients. This data base includes consecutive admissions to the adult ICUs of four hospitals in the northeastern United States. The second is a similar data base from the European-North American Study of Severity Systems (ENAS), and it includes 20 hospitals. The third data base, a 1991 national survey of ICUs by the Society of Critical Care Medicine (SCCM), lists characteristics of patients in ICUs in the United States on a specific day. Logistic regression was used to analyze the first two data bases; the percentage of patients in New York with DNR orders was calculated for each of the three data bases and compared with patients in neighboring states. Length of ICU and hospital stay was measured in the first two data sets. In the MPM data, 14.4% of medical patients in New York had a DNR order written at the time of ICU discharge, compared with 198% of medical patients in Massachusetts; and 4.3% of New York surgical patients had a DNR order written at the time of ICU discharge, compared with 8.3% of surgical patients in Massachusetts. In the ENAS data, 7.4% of New York nonoperative patients has a DNR order in place within 24 hours, compared with 8.4% of such patients in the other states; and 1.0% of New York operative patients had DNR orders, compared with 3–5% of operative patients from other states. Logistic regression revealed that a New York patient was less likely to have a DNR order written than a patient located in one of the other states studied. Data from the SCCM survey demonstrated that the New York percentage of patients with “no CPR” orders was 5.50%, compared with a percentage of 6.87% in other states. With few exceptions, these differences between New York and surrounding states did not have an impact on hospital length of stay. During the period studied following implementation of New York's DNR Law, utilization of DNR orders in New York State was significantly lower than neighboring states. This decreased utilization, however, did not effect hospital utilization as measured through length of stay and ICU admissions.