We audited and analysed the adverse effects and safety of postoperative pain management on 2509 consecutive patients under care of the Acute Pain Service at a tertiary referral teaching hospital over a 32-month period. Our standard respiratory monitoring consisted of continuous pulse oximetry, hourly respiratory rate counting, sedation scoring and intermittent arterial blood gas sampling. This protocol was reliable and detected six episodes of bradypnoea, 13 of hypercapnia and 23 of oxygen desaturation occurring in 39 patients (1.8% of all spontaneously breathing patients). Two patients required naloxone injection and none had long-term sequelae. Hypotension due to epidural bupivacaine 0.0625% and fentanyl 3.3 micrograms.ml-1 infusion occurred in four patients (1.2%), all with a sensory block higher than T5. They readily responded to fluid infusion and ephedrine (two patients). Postoperative nausea or vomiting occurred in 723 (28.8%) and 380 (15.1%) patients, respectively. Odds ratio analysis showed that the risk factors for postoperative nausea and vomiting were: female gender, gynaecological operations, nongeriatric patients and systemic analgesia. Postoperative nausea and vomiting decreased analgesic efficacy by discouraging the use of patient-controlled analgesia and was regarded as equally distressing as pain. Other side-effects included: pruritus in 182 patients; dizziness in 333 and lower limb weakness in 73 (21.2% of patients receiving epidural local anaesthetics). It is concluded that a standard monitoring and management protocol, an experienced nursing team and reliable Acute Pain Service coverage is mandatory for the safe use of modern analgesic techniques.