A randomised crossover study was undertaken to compare the quality and cost of controlled versus empirical sedation with midazolam in critically ill patients. Patients (n = 40) entering the ICU were enrolled provided they satisfied the strict entry criteria. During 90 hours of midazolam sedation, patients received randomly allocated 10-hour periods of controlled or empirical sedation. With empirical sedation, the mean dose of midazolam and the cost of sedation were almost double those with controlled sedation. The quality of sedation was superior with the controlled method. In a separate study on 352 patients, a cost-benefit analysis of controlled sedation with midazolam or propofol infusion or bolus injections of morphine plus diazepam showed that the quality of sedation achieved with propofol was superior to the other two regimens and that, with morphine plus diazepam, the quality of sedation was unacceptably poor. Although the direct purchase price of propofol was higher than that of other agents, the total cost of sedation with propofol was lower than that for midazolam for short-term intensive care (less than 24 hours) and comparable to midazolam for longer-term use. However, indirect benefits of sedation with propofol include a much shorter ICU stay with the attendant reduced nursing costs and greater throughout the patients, and this more than compensates for the higher purchase price of the agent.