Question: Does the common practice of infusing small amounts of glucose after cardiopulmonary arrest worsen neurologic outcome? Design and setting: A community-based randomized trial in Seattle, WA. Paramedics treated all patients with out-of-hospital cardiac arrest in a standard fashion except that the intravenous infusion did or did not contain glucose; ie, patients received either usual treatment, with 5% dextrose in water (D5W), or alternative, with half normal saline (0.45S). Outcomes: The main outcome was awakening, defined as the patient having comprehensible speech or following commands as determined by chart review. Other outcomes were survival to hospital admission and to discharge. Results: Over 2 years, paramedics randomized 748 patients. The type of fluid administered was not significantly related to awakening (16.7% for D5W versus 14.6% for 0.45S), admission (38.0% for D5W versus 39.8% for 0.45S), or discharge (15.1% for D5W versus 13.3% for 0.45S). As in previous studies, patients whose arrest had likely been on a cardiac basis with initial rhythms of ventricular fibrillation or asystole had admission blood glucose levels significantly related to awakening: mean = 309 mg/dl for never awakening and 251 mg/dl for awakening. Of note, the relation between glucose and awakening was reversed in the remaining patients, who had electromechanical dissociation or noncardiac mechanisms of arrest. Conclusion: Current practices of using limited amounts of glucose-containing solutions after cardiopulmonary arrest do not need to be changed. Blood glucose level on admission is a prognostic indicator but depends on the type of arrest.