Propofol and midazolam are often used for sedation in the intensive care unit. The aim of this systematic review was to estimate the efficacy and harm of propofol versus midazolam in mechanically ventilated patients. A systematic search (Medline, Cochrane Library, Embase, bibliographies), any language, up to June 1999 was performed for reports of randomized comparisons of propofol with midazolam. Data from 27 trials (1624 adults) were analyzed. The average duration of sedation varied between 4 and 339 h. In 10 trials, the duration of adequate sedation was longer with propofol (weighted mean difference 2.9 h; 95% confidence interval [CI], 0.2–5.6 h). In 13 trials (mostly postoperative), sedation lasted 4 to 35 h; in 9 of those, average weaning time from mechanical ventilation with propofol was 0.8–4.3 h; with midazolam it was 1.5–7.2 h (weighted mean difference 2.2 h [95% CI, 0.8 to 3.7 h]). In 8 trials, sedation lasted 54 to 339 h; there was a lack of evidence for difference in weaning times. Arterial hypotension (relative risk 2.5 [95% CI, 1.3 to 4.5]; number-needed-to-treat, 12), and hypertriglyceridemia (relative risk 12.1 [95%CI, 2.9 to 49.7]; number-needed-to-treat, 6) occurred more often with propofol. The duration of adequate sedation time is longer with propofol compared with midazolam. In postoperative patients with sedation <36 h, weaning is faster with propofol.