This observational study aimed to identify simple electroencephalogram indices of inadequate intraoperative opioid-mediated nociceptive blockade and to compare these indices with routinely used clinical predictors of severe postoperative pain in adults. Intraoperative trend and waveform data (electrocardiogram, pulse oximetry and electroencephalogram) were collected, pain intensity in the post-anaesthesia care unit was quantified using an 11-point Verbal Rating Score, and opioid administration was recorded. Using the initial post-anaesthesia care unit Verbal Rating Score as the primary endpoint, the relationship between five possible explanatory variables--surgery type, depth of volatile anaesthesia (minimum alveolar concentration), electroencephalogram signs (state entropy, spindle-like activity and delta-band power) and estimated end-of-operation effect-site morphine concentrations--was examined. One hundred and thirteen patients were recruited, with 94 included in the final clinical and electroencephalogram data analysis. Fifty-two patients had moderate or severe pain (Verbal Rating Score > or = 5). State entropy was lower (46.5 +/- 2.9 vs 43.1 +/- 1.9, P = 0.04) and spindle-like activity higher (0.42 +/- 0.03 vs 0.50 +/- 0.02, P = 0.03) in the moderate/severe pain group. [corrected] These findings suggest that there is a modest association between electroencephalogram measures near the end of surgery and the severity of postoperative pain.