Several not conventional medicine treatments aspirate to legislative acknowledgments of full integration in the health system. On the other hand, registered drugs not always produce the promised effects and, often, constitute an additional cause of dead. Moving from recent debates on the therapeutic effectiveness, this article describes and analyses those aspects of the medical-scientific methodology that assert, in probabilistic terms, the health alterations due to properties of administered drugs and not to chance. More particularly, as in the last years, the planning of the clinical trials underwent substantial changes, mostly for economic reasons, it examines the differences among superiority, equivalence and not inferiority studies highlighting the different logic. As general consideration, the former kind of study is to be preferred while the other two can be accepted only when the equivalence margin is very close.