The objectives of this study were to describe the sedative, neuromuscular blocking agents (NMBA) and reversal agents utilized in adult intensive care units across the United States and determine the adherence to American College of Critical Care Medicine and Society of Critical Care Medicine (SCCM) guidelines. In addition, the authors assessed the use of written protocols, criteria used for selecting these agents, and monitoring practices. Questionnaires were mailed to attending physician members of SCCM in the spring of 1998. A cover letter was enclosed that explained the purpose of the survey asking the respondent to forward the questionnaire to a colleague if unable to complete. Four-hundred fifty-seven questionnaires were returned representing 393 different institutions for a response rate of 50.4% (393/780). Respondents were physicians (91.2%) practicing in a community (49.7%) or university teaching hospital (38.3%). The sedative agents used most often were opioids and benzodiazepines for >72 hours, and NMBA utilized were vecuronium and pancuronium for >24 hours. The most often cited indications for sedatives were agitation, anxiety/fear, and facilitation of intubation and maintenance of mechanical ventilation for NMBA. Only 32.6% used written protocols for sedatives and 46.8% for NMBA. Decisions regarding agent selection were based on clinician preference and experience and agent duration of action. Seventy-eight percent monitored sedative use primarily with the Glasgow Coma Scale and the modified Ramsay score. Monitoring of NMBA was used more frequently (91.3%) with peripheral nerve stimulation. The most common reversal agents used were naloxone and flumazenil for adverse drug effects. While many of the respondents indicated they used morphine and lorazepam for long-term sedation, the majority utilized midazolam and propofol for >24 hours despite the recommendation of SCCM. Vecuronium was prescribed more routinely than pancuronium. The number of institutions utilizing protocols for any of these agents was low; instead, decisions were based on clinician preference.