Animal screening tests and methods employing induced pain in human volunteers have failed to predict with any consistency the clinical performance of analgesic drugs. These drugs must be assayed in a clinical setting in which they might be used therapeutically. The lack of adequate objective standards for the measurement of pain or pain relief, the variability in responses among patients, and the influence of environmental factors on these responses, all make it essential that the study be adequately controlled and so designed as to provide some measure of the sensitivity of the method and of the statistical significance of the results.Such a method for evaluating analgesics in patients with chronic pain due to cancer has been employed at the Memorial Cancer Center for the past 8 years. Measurements of analgesia are derived from the patients' own estimates of pain intensity both before and at hourly intervals after a test medication Yielding data which allow comparison of drugs in terms of peak and total effects. Using these criteria, studies of the effects of morphine and a placebo were carried out and provide the basis for a method of assaying other analgesic drugs. A prototype of the latter is presented in the form of a study of aspirin, morphine, and a combination of these drugs.