Fentanyl (1 microgram/kg body weight) was administered intravenously and via a lumbar epidural catheter (in random order) on 2 separate occasions to 6 patients with chronic pain associated with non-terminal disease states. Frequent blood samples were collected from an indwelling intravenous catheter and CSF samples were collected via spinal needles inserted in the cervical (C7-T1 interspace) and lumbar (L3.4 interspace) regions at 0, 5, 10, 20, 30 and 45 min after fentanyl administration. The concentration of fentanyl in blood and CSF samples were quantified by a sensitive and selective gas-liquid chromatography assay. Visual analogue pain scores (VAPS) were recorded every 5 min for the first hour. Coded syringes (one containing the appropriate fentanyl dose while the other contained an equivalent volume of saline) allowed the investigator administering the fentanyl and assessing VAPS to remain blinded as to which route of administration actually contained the fentanyl. There was minimal vascular uptake of fentanyl following epidural administration. Similarly, the permeation of fentanyl into cervical and lumbar CSF following intravenous administration was minimal and erratic such that only 4 of the 60 CSF samples had detectable fentanyl concentrations. In contrast, there was a rapid penetration of fentanyl across the dura mater following lumbar epidural administration. There was significantly fentanyl in lumbar CSF samples by 10 min in 5 patients, and by 20 min in all 6 patients. The mean maximum lumbar CSF concentration was 19.1 ng/ml, while the time associated with these maximum concentrations was 22.5 min. The mean maximum cervical CSF fentanyl concentrations were 10% of the lumbar CSF concentrations.(ABSTRACT TRUNCATED AT 250 WORDS)