More than 80 deaths have occurred after the use of midazolam (Versed), often in combination with opioids, to sedate patients undergoing various medical and surgical procedures. We investigated the respiratory effects of midazolam (0.05 mg.kg-1) and fentanyl (2.0 micrograms.kg-1) in volunteers. The incidence of hypoxemia (oxyhemoglobin saturation less than 90%) and apnea (no spontaneous respiratory effort for 15 s) and the ventilatory response to carbon dioxide were evaluated. Midazolam alone produced no significant respiratory effects. Fentanyl alone produced hypoxemia in half of the subjects and significant depression of the ventilatory response to CO2, but did not produce apnea. Midazolam and fentanyl in combination significantly increased the incidence of hypoxemia (11 of 12 subjects) and apnea (6 of 12 subjects), but did not depress the ventilatory response to CO2 more than did fentanyl alone. Adverse reactions linked to midazolam and reported to the Department of Health and Human Services highlight apnea- and hypoxia-related problems as among the most frequent adverse reactions. Seventy-eight per cent of the deaths associated with midazolam were respiratory in nature, and in 57% an opioid had also been administered. All but three of the deaths associated with the use of midazolam occurred in patients unattended by anesthesia personnel. We conclude that combining midazolam with fentanyl or other opioids produces a potent drug interaction that places patients at a high risk for hypoxemia and apnea. Adequate precautions, including monitoring of patient oxygenation with pulse oximetry, the administration of supplemental oxygen, and the availability of persons skilled in airway management are recommended when benzodiazepines are administered in combination with opioids.