Midazolam is a new water-soluble benzodiazepine with a much shorter pharmacologic half-life than diazepam. Despite this shorter pharmacologic half-life, several reports indicate that patients do not recover more rapidly after sedation with midazolam than with diazepam. The purpose of this study was to compare recovery of patients sedated with either midazolam or diazepam alone or in combination with fentanyl using the digit symbol substitution test (DSST) and Trieger test. Patients were randomly divided into treatment groups and recovery tests were administered to the patients prior to sedation and at 60, 120, and 180 minutes after achieving a standardized sedative endpoint. Patients who received midazolam alone had significantly fewer numbers of correct reponses on the DSST than patients who received midazolam plus fentanyl or diazepam with or without fentanyl. When midazolam was combined with fentanyl there was no significant difference between results obtained on the DSST when compared with either diazepam group. Comparisons between all groups using dots missed or millimeter deviation on the Trieger test showed no statistical difference between any groups. These data indicate that midazolam as a single IV agent has a slightly prolonged recovery phase compared to diazepam. The addition of fentanyl to the sedation regimen allows reduction in the midazolam dose resulting in a recovery time comparable to that of diazepam.