Controlled studies combining the resources of several hospitals and investigators with carefully designed prospective protocols have become increasingly important in the evaluation of various modes of medical and surgical therapy. In the past, therapeutic practice has been determined largely by methods of trial and error or vague "clinical impressions," at best imprecise and difficult to measure. The use of historical controls or comparisons based on dissimilar groups of patients, while somewhat better, still invites erroneous and hazardous conclusions especially when buttressed by large but noncomparable numbers. When the therapeutic effect is great or the difference between therapies being compared is large, these methods have been satisfactory, although medical history is replete with many brilliant innovations which were accepted only with great reluctance by begrudging colleagues. When the therapeutic difference is small more precise methods of measurement are essential to reach valid conclusions. In some regards controlled studies may serve an