BACKGROUND To evaluate the postoperative analgesic effect of epidural morphine administered at different timing in lumbar spine surgery. METHODS Eighty-four patients who were scheduled for elective lumbar spine surgery were randomized in three groups. Seventeen patients in group I who received non-steroid analgesics postoperatively (diclophenac sodium 50 mg, iv, q4h) served as control while thirty-six patients in group II who received single dose epidural morphine 3 mg in combination with 10 ml 2% lidocaine given at the lesion site (L4-5 or L5-S1) just before general anesthesia and thirty-one patients in group III who received 3 mg morphine in combination with 3 ml 2% lidocaine administered to the targeted epidural space by means of slow drippings just before wound closure were studied subjects. RESULTS During the first 24 h postoperatively, the patients in group II and group III suffered a pain which was significantly less in intensity as compared with those in group I (p < 0.05). We used the 10 cm visual analog pain score (VAS) to scale post-operative pain with "no pain" and "worst pain" respectively anchored at 0 and 10 cm. The incidence of side effects such as pruritus, nausea and vomiting was higher in group II and III than in group I. We did not evaluate the occurrence of urinary retention because routine retention urinary catheterization in all patients hampered us to do so. There were no significant differences in the quality and duration of analgesia between group II and III. Respiratory depression of clinical significance was not observed. Neither decrease in oxygen saturation below 92% registered on pulse oximetry nor decrease in respiratory rate below 12 cycles/min was found in the PACU. CONCLUSIONS Preoperative or intraoperative administration of epidural morphine could provide satisfactory analgesia in lumbar spine surgery during the first 24 h postoperatively.