Although epidural opioids frequently are used to provide postoperative analgesia, several articles have suggested that the analgesia after epidural fentanyl is similar to that after an equal dose of fentanyl given intravenously. To address this issue further, 29 postthoracotomy patients were studied in a randomized, double-blinded trial comparing a lumbar epidural fentanyl infusion with an intravenous fentanyl infusion for analgesia, plasma fentanyl pharmacokinetics, and respiratory effects for 20 h postoperatively. In all patients in both groups, good analgesia was achieved (pain score less than 3, maximum 10) over a similar time course, although the patients receiving epidural infusion required a significantly larger fentanyl infusion dose than did the patients receiving intravenous infusion (group receiving epidural fentanyl infusion: 1.95 +/- 0.45 micrograms.kg-1.h-1; group receiving intravenous fentanyl infusion: 1.56 +/- 0.36 micrograms.kg-1.h-1; P = 0.0002). The time course for the plasma fentanyl concentrations was similar in the two groups, and plasma fentanyl concentrations were not significantly different at any sampling period (T7-T20; group receiving epidural fentanyl infusion: 1.8 +/- 0.5 ng/ml; group receiving intravenous fentanyl infusion: 1.6 +/- 0.6 ng/ml; P = 0.06). Similarly, calculated clearance values for the two groups were not significantly different (group receiving epidural fentanyl infusion: 0.95 +/- 0.26 l.kg-1.h-1; group receiving intravenous fentanyl infusion: 0.87 +/- 0.25 l.kg-1.h-1; P = 0.3). Both groups demonstrated a similar degree of mild to moderate respiratory depression postoperatively, which was assessed with continuous respiratory inductance plethysmography and sequential arterial blood gas analysis. Side effects (nausea, vomiting, pruritus) were mild and did not differ between groups.(ABSTRACT TRUNCATED AT 250 WORDS)