This paper discusses some of the problems involved in drawing conclusions across studies about the efficacy of biofeedback. It focuses on biofeedback for the treatment of hypertension, but the same difficulties arise when considering the effect of biofeedback in other disorders. Large multicenter studies using the same inclusion and exclusion criteria, biofeedback protocol, and methodology are badly needed if biofeedback practitioners are ever going to demonstrate the real effectiveness of biofeedback.