Our previous study suggested that when compared between patients, morphine, pethidine and fentanyl were equally satisfactory for use in patient-controlled analgesia (PCA), although quantitative differences in their side-effect profiles were detectable. The present study evaluated whether individual patients could detect differences or express preferences for individual opioids when treated by PCA with all three in random sequence finishing with the first administered opioid. The main side effects were pruritus, nausea and vomiting. There were few differences in patients' responses to morphine, pethidine and fentanyl, or of satisfaction with these drugs, across patients, but individual patients' responses to the opioids could be highly variable. Some patients were able to tolerate all three opioids investigated, some were intolerant to all and some patients appeared to be sensitive to one or two of the opioids but show a preference for the remainder. These findings support the clinical practice of changing from one opioid to another (with good effect) in post-operative patients experiencing intolerable side-effects. The reasons for a patient responding differently to different opioids and even to the same opioid on separate occasions are not clear and appear inexplicable on the grounds of currently postulated receptor affinities or of physicochemical properties of the opioids studied. A plethora of factors will influence how an individual patient will respond to surgery and how he/she will recover. The physiological response to opioids is one variable which appears to be influenced by this complex set of factors and in turn will affect them. The findings of this study, like that of its predecessor, suggest that morphine, pethidine and fentanyl can be used successfully in PCA and that for some patients who are responding poorly, changing the opioid may be beneficial.