Symptoms of obstructive sleep apnoea are common in patients presenting for surgery and are associated with increased morbidity. Analgesia contributes significantly to postoperative respiratory depression and obstruction, so we compared standard morphine patient-controlled analgesia with an opioid-sparing protocol (tramadol patient-controlled analgesia, parecoxib and rescue-only morphine) in these patients. Sixty-two patients presenting for elective surgery with body mass index > or = 28 and signs or symptoms suggesting obstructive sleep apnoea were randomised to receive either the opioid or opioid-sparing postoperative analgesia protocol, with continuous respiratory monitoring for 12 hours on the first postoperative night. The number of respiratory events (apnoeas and hypopnoeas) and oxygen desaturations were compared. There was no difference between treatment groups in the number of obstructive apnoeas, hypopnoeas or central apnoeas. However, central apnoeas and a rate of respiratory events > 15 per hour were related to postoperative morphine dose (P = 0.005 and P = 0.002). In patients at risk of obstructed breathing, intention to treat with an opioid-sparing analgesia protocol did not decrease the rate of respiratory events, although the rate was still related to the total morphine dose.