True improvements in the treatment of cancer-by the introduction of newdrugs or novel drug combinations, new therapeutic modalities, ortechnologic improvements of old modalities-result in higher response ratesand prolonged survival when compared with existing therapies. When a newtreatment convincingly meets the test of improving survival rates or, atworst, improving patients' quality of life, it becomes the acceptedstandard of care if its side effects are acceptable and its cost is notprohibitive. Improved therapeutic results can be demonstrated only byclinical trials with an adequate numbers of patients, appropriate controlsubjects, and a sufficient duration of follow-up. Therapeutic breakthroughsare revolutionary advances in treatment, usually rapidly and dramaticallyobvious in comparison with historic controls; demonstration of benefit inthese cases does not usually require randomized trials. Much more common, however, are new therapies that represent modest, incremental advances overexisting treatment and that usually require randomized comparison trials todemonstrate convincingly statistically significant improvement. Arandomized clinical trial should test an important hypothesis. It must becarefully designed to ensure that both groups of patients are comparable interms of various prognostic variables and to minimize subtle sources ofbias. An honest belief that both arms of the trial are a priori equal mustbe maintained. Meeting these criteria, the randomized clinical trial offersto the individual cancer patient treatment that should be at least equal tothe best available nonexperimental therapy. This equates with Good Medicine.