Thirteen post-thoracotomy patients were entered into a double-blind, randomized clinical trial comparing the effects of epidural morphine (Group E) and intravenous morphine (Group I) on postoperative respiratory depression. Postoperative respiratory depression was assessed for 24 hours by (a) PaCO2 at 2, 6, 12 and 24 hours (b) hourly assessment of respiratory rate (RR) (c) presence of respiratory rate of less than ten breaths per min for greater than 5 min (SRR) (d) hypopnoea/apnoea (H/A). RR, SRR, and H/A were measured using respiratory inductive plethysmography. PaCO2 was significantly elevated at 2, 6 and 12 hours in Group E and only at two hours in Group I. One of five patients in Group I had a single episode of SRR whereas five of eight patients in Group E had multiple episodes of SRR. None of the patients in Group I had H/A episodes, in contrast to six of eight in Group E who had numerous H/A episodes post-operatively. This difference was statistically significant. Multiple doses of epidural morphine produce an insidious and unpredictable change in respiratory pattern. Electronic monitoring is useful to assess those at risk of overdose and possible respiratory arrest.