The cohort for this study included 39 patients, between the ages of 8 and 20 years, who had had thoracotomy. Postoperative analgesia was provided by one of three techniques: intravenous narcotics (20 patients), thoracic epidural catheter (10 patients), or interpleural analgesia (IPA) (nine patients). Both IPA and epidural analgesia were administered according to a specific protocol. The efficacy of the three methods was evaluated using two indicators: the total intravenous narcotic requirements for the initial 72 hours and the number of times a dose of intravenous narcotic or supplemental epidural fentanyl was administered to each patient. Patients in the IPA group required significantly less intravenous narcotics (P less than .05) during the first three postoperative days (2.2 +/- 0.4, 1.9 +/- 0.6, and 1.4 +/- 0.5 mg of meperidine/kg/day) than patients in the intravenous narcotic group (8.1 +/- 1.2, 7.2 +/- 0.9, and 5.6 +/- 1.2 mg of meperidine/kg/day). When comparing epidural analgesia and IPA, the number of interventions in the patients receiving epidural analgesia was significantly less (P less than .05) than in the group receiving IPA. Four of 10 patients in the epidural group and two of nine in the IPA group required no interventions during the initial 72-hour postoperative course. In the epidural analgesia group, there were 19 days (of a total of 30 days) during which patients required no interventions. This percentage was significantly greater (P less than .05) than that of the IPA group, in which there were 9 days (of a total of 27 days), during which no interventions were required.