Objective:To test the following hypotheses: there is a correlation between spread of epidural analgesia as assessed postoperatively by pinprick/cold test and postoperative pain intensity; block of pinprick/cold sensation is associated with absence of postoperative pain.Design:Correlation analysis on prospectively collected data.Setting:University hospital.Patients:One hundred patients undergoing major surgery. Consecutive sample.Interventions:Patients received an epidural infusion of bupivacaine 1 mg/ml, fentanyl 2 μg/ml, and epinephrine 2 μg/ml for at least 48 hours postoperatively. The infusion rate was adjusted according to pain intensity, occurrence of hypotension, or motor block.Outcome Measures:Assessments were made on three time points: 20-24 hours, 32-36 hours, and 42-48 hours after extubation. Assessments included pinprick and cold sensitivity from C2 to S5, pain intensity (visual analogue scale, VAS) at rest, after cough, and after mobilization. Data were analyzed by multiple regression.Results:VAS significantly decreased with increasing spread (number of dermatomes for which hyposensitivity to pinprick or cold was observed). Spread could explain only 2-5% of the variability of VAS. Absence of both pinprick and cold sensation at all dermatomes corresponding to the surgical wound was frequently associated with pain. A high proportion of patients manifesting an upper level of block above T5 had pain after abdominal surgery.Conclusions:Spread and efficacy of epidural analgesia as assessed by pinprick and cold stimulation correlate poorly with postoperative pain. These methods are of limited value both as clinical indicators of the efficacy of postoperative pain control and for investigating the effect of epidural drugs and techniques.