Twenty-five patients with pain due to advanced cancer were randomized to receive patient-controlled so injections (PCLS) of Hydromorphone (HM) versus continuous so infusion (CSCI) of HM by means of a Pharmacla 5200 Pump. Each self-Infusion (CSCI) of HM by means of a Pharmacia 5200 pump. Each self-injection of HM during Pci was equlavent to 4 hours of CSCI After 3days a crossover occurred, and patients received the alternate treatment for 3 days. During both phases of the study, patients could request extra doses of HM from their nurses. In 22 patients able to be evaluated, pain intensity (visual analogue, o-100mm) at 9.00 a.m.and 4.00 p.m. was 31±23 and 28±18 and 27± on PCI versus 28 ± 18 and 27± respectively on PCI versus 23±18 and 27 ±17mm, respectively, on CSCI (p=not significant (NS), The total dose of Hm was 168±197 and 181±234 mg on PCI and CSCI, respectively 9p=Ns), No significant difference was found in nausea, drowiness, or number of hours of sleep. The total number of extra doses of Hm was 6±7 on CSCI versus 2±3 on PCI (p=.007). At the end of the study, patients chose PCI and CSCI in seven and 10 case, respectively (p=NS; 55 patients expressed no preference), We conclude the both methods were similar in re-gard to effectiveness and taxic effects in short-term hospital use. [J Natl cancer Inst 1988;801152–1154]