A retrospective investigation was undertaken of 48 cancer patients receiving long-term treatment with high dosages of epidural opioids. Prior to introduction of the epidural catheter, the patients were in stable oral opioid treatment (median dose: 300 mg morphine/24 hours, range 40-1,360). The effect of stable epidural opioid treatment (median dose: 90 mg morphine/24 hours, range 16-600) in the first catheter was assessed as good in 20, reasonable in 14 and poor in 13. The effect could not be assessed in one patient. Twenty-nine patients could be assessed on pure and stable epidural opioid treatment. The effect was good in 14, reasonable in seven and poor in eight. If the patients were subdivided according to the type of pain, 16 had somatic and/or visceral types of pain while 13 had neurogenic pain. Epidural opioid treatment of neurogenic pain was found to produce statistically significantly poorer results than in the case of somatic/visceral pain (p less than 0.05). At the time of assessment, no statistically significant difference was present in the epidural opioid dosage between the groups. The patients were subdivided into a low dosage group and a high dosage group. The low dosage group comprised patients who had received oral opioid treatment with less than 300 mg morphine/24 hours and the high dosage group comprised patients who received equal or greater than 300 mg morphine/24 hours. Epidural opioid treatment was found to provide statistically significantly better results in the low dosage group (p less than 0.05). In this group, a tendency to better relief of neurogenic pain was observed as compared with the high dosage group. At the time of assessment, no statistically significant difference was present in the epidural opioid dosage between the high dosage and low dosage groups.(ABSTRACT TRUNCATED AT 250 WORDS)