We examined the effect on the quality of analgesia and side effects of increasing the patient control component of morphine patient-controlled analgesia (PCA) by offering the patient a choice of bolus dose sizes. Using a three-button hand piece, patients could choose between 0.5-, 1.0-, and 1.5-mg boluses of morphine (variable-dose PCA, VDPCA). Successful demands were delivered by a modified Graseby 3400 Anaesthesia Pump controlled by a Toshiba T1900 computer. This system was compared with conventional fixed-dose PCA (FDPCA) (1.0 mg of morphine) delivered by a Graseby 3300 PCA Pump. Both treatment groups had a 5-min lockout interval. Sixty patients were randomly assigned to receive either VDPCA or FDPCA after major abdominal gynecological surgery or hip or knee arthroplasty. Treatment groups did not differ in their duration of PCA therapy, total morphine consumption, or time spent with mild or severe oxyhemoglobin desaturation. There were no differences in their ease of controlling pain, satisfaction with pain control, experience of pain on movement, quality of sleep, severity of nausea, or incidence of vomiting. Although the more complex VDPCA technique provides adequate postoperative analgesia, it does not offer any advantage over conventional FDPCA.