The purpose of this study was to investigate whether regression of sensory analgesia during constant epidural bupivacaine infusion was different in postoperative patients with acute pain than in patients with chronic nonsurgical pain. Sensory levels of analgesia (to pinprick) and pain (on a five-point scale) were assessed hourly for 16 hours during continuous epidural infusion of 0.5% plain bupivacaine (8 ml/hr) in 12 patients with chronic nonsurgical pain and in 30 patients after major abdominal surgery performed under combined bupivacaine and halothane--N2O general anesthesia. No opiates were given. If sensory analgesia decreased more than five segments from the initial level or if the pain score reached 2 (moderate pain), the patient was removed from the study. Initial levels of sensory analgesia after loading doses of 21.8 +/- 0.5 and 19.3 +/- 0.8 ml bupivacaine 0.5% were similar (T3.8 +/- 0.3 and T3.8 +/- 0.5) in the surgical and chronic pain patients, respectively (mean +/- SEM). Of the surgical patients, only 4 of the 30 (13%) maintained the initial level of sensory analgesia, and a pain score below 2 throughout the study compared with 7 of the 12 patients with chronic pain (58%) (P less than 0.01). Mean duration of sensory blockade was significantly longer (P less than 0.005) in the patients with chronic pain than in surgical patients (13.1 +/- 1.2 and 8.5 +/- 0.7 hours, respectively). Thus, surgical injury hastens regression of sensory analgesia during continuous epidural bupivacaine infusion. The underlying mechanism remains to be determined.