To study the efficacy and safety of continuously administered intravenous morphine for cancer pain unrelieved by standard narcotic therapy, bolus intravenous injections of 2 to 5 mg of morphine were given every 10 minutes until pain relief was achieved. Within the next hour, continuous intravenous morphine infusion was begun with the hourly dose equal to the cumulative bolus dose. Respiratory rate, pulse, blood pressure, arterial blood gas values, mental status, and pain relief were recorded at baseline and during the study period. A reduction in arterial oxygen pressure (PaO2) and/or increase in arterial carbon dioxide pressure PaCO2 of more than 20 percent of baseline values occurred, during the first 24 hours of infusion, in a minority of patients. This did not require changes in hourly morphine dose. Despite subsequent increases in morphine dose, blood gas values tended to remain at or return toward baseline values. Severe toxicity occurred during one trial and was heralded by bradypnea and marked somnolence. Major pain relief was achieved in 11 of 15 trials. Therefore, continuous intravenous morphine is effective and safe therapy. Bradypnea associated with marked somnolence is a cause for dose reduction.