For approval of new drugs, the law requires a minimum of two independent well-controlled studies for each indication. The preferred study design is prospective, randomized, and comparative. The use of historical controls, while sometimes acceptable, must be justified. A favorable effect on survival and/or quality of life is generally required for approval. The study of cancer drugs in combination presents problems which are discussed. Drug firms are strongly urged to discuss the protocols with the Food and Drug Administration prior to initiating phase III comparative studies to be used for approval of the drug.