Patient-controlled analgesia (PCA, intravenous self-application of narcotics) was studied during the early postoperative period. Subjects were 40 ASA I-III patients recovering from elective major and minor surgery (each 20 having undergone abdominal or orthopaedic operations). Pentazocine bolusses of each 8 mg were available via a hand-button whenever the patients felt pain relief necessary, and delivered by a microprocessor-controlled injection pump (On-Demand Analgesia Computer, ODAC). Hourly maximum dose was set to 60 mg with a pump refractory time of 1 min between valid demands. A continuous low-dose pentazocine infusion (1 mg/h) was additionally administered in order to prevent catheter obstruction. Duration of the PCA period was 20.3 +/- 5.9 h (mean, standard deviation). During this time, 20.0 +/- 12.7 demands per patient were recorded resulting in mean pentazocine consumption of 135.6 +/- 81.4 micrograms/kg/h. Self-administration was characterized by considerable intra- and interindividual variability. There were no statistically significant differences with regard of pentazocine consumption or pain relief between abdominal and orthopaedic patients, nor could any be demonstrated between the sexes. Similarly, no clear differences were found after various anaesthetic techniques (neuroleptanalgesia, halothane or spinal anaesthesia). Over-all efficacy and patient acceptance proved to be excellent. Effectiveness of PCA was judged superior by about 68% of patients when compared with previously experienced conventional postoperative analgesia. Side effects (nausea, emesis, sweating) occurred in about 10-18% but were usually of minor intensity. Circulatory or respiratory problems were not observed during the PCA period. Patient-controlled analgesia is discussed as a promising concept for the treatment of acute pain and clinical pain research.