The authors conducted a randomized double-blind controlled study comparing groups of patients receiving iv or epidural fentanyl infusions to determine whether, at comparable levels of analgesia, 1) the severity of side effects was different; and 2) plasma fentanyl concentrations differed between the two groups. Twenty-eight ASA physical status 2 women scheduled to undergo elective cesarean section were randomized into two groups to either receive fentanyl intravenously and saline epidurally or fentanyl epidurally and saline intravenously. After delivery of the infants under epidural anesthesia, each patient received a bolus of fentanyl 1.5 microgram/kg either intravenously or epidurally, and a fentanyl infusion was begun via the same route. Concurrently, a saline bolus and infusion were given via the alternate route. The rates of the fentanyl and saline infusions were adjusted until each patient was comfortable. Patients rated their pain, nausea, and pruritus on visual analogue scales. Sedation was evaluated by an observer. Respiratory depression was evaluated by end-tidal PCO2. Data were analyzed by unpaired two-tail t tests. Plasma fentanyl concentrations were measured at 12 and 24 h. Three patients in the iv group were dropped from the study because of inadequate pain relief. For the remaining 25 patients, similar infusion rates of fentanyl were required to produce similar levels of analgesia at 12 and 24 h. The severity of nausea, pruritus and sedation, and end-tidal PCO2 were similar for both groups. The plasma concentrations of fentanyl were significantly greater in those who received iv fentanyl at 12 h but not at 24 h.(ABSTRACT TRUNCATED AT 250 WORDS)