This study examined midazolam and midazolam plus fentanyl in a placebo-controlled, double-blind clinical trial. It tested the hypothesis that combined drug therapy results in significantly poorer safety but no difference in efficacy compared to the single drug approach. Subjects were among 207 mildly anxious young adults having their third molars removed. Fentanyl had a significant depressant effect on respiration. Fifty of 79 (63%) subjects who received a midazolam-fentanyl combination became apneic, while only two of 78 (3%) who received midazolam alone were apneic (Fisher's Exact Test, P < 0.001). Two subjects (2.5%) in the combination group and none in the midazolam alone group had oxygen saturations drop below 90%. About twice as many subjects in the combination group had end-tidal carbon dioxide (EtCO2) levels greater than 25% above baseline. While these results are consistent with those for apnea, contingency analyses of the oxygen saturation and EtCO2 results were not statistically significant. Subjects in the combination group were more than four times as likely to have excellent versus good, fair, or poor sedation at a given level of intraoperative pain, and behavioral (movement and verbalization) but not cognitive measures of anxiety were attenuated.