Limb ischaemia induced by a sub-maximum effort tourniquet technique was used to characterize the analgesic effects of lumbar epidural morphine in volunteers. As an index of pain threshold, we measured the time to perception of pain in and upper an a lower limb before and at intervals up to six hours following epidural injections of morphine 3.5 mg an 7.0 mg, and before and after subcutaneous injections of the same doses. Subcutaneous morphine had no significant effect on the times to perception of pain in either limb. Lumbar epidural morphine did not alter upper limb times, but markedly delayed the onset of pain in the lower limbs. This lower limb analgesic effect was apparent thirty minutes after injection, peaked at about ninety minutes and was still present after six hours. Serum levels of morphine were nearly identical after subcutaneous and epidural injections of the same dose. We conclude that lumbar epidural morphine produces marked analgesia for this type of experimental pain primarily by a "regional" effect rather than as a result of systemic absorption. This regional effect develops slowly and is prolonged.