There is currently a lively debate involving scholars from diverse cultural background on the subject of evidence-based medicine. In order to set out the grounds of discussion this paper deals with the problem of "evidence", both the concept and meaning of this term. After a brief history of the idea of "evidence" from antiquity to the present day, it will be underlined how modern day thinking recognizes the existence of two types of "evidence": rational evidence supported first by René Descartes, and perceptible evidence, defended by Francis Bacon. In contemporary epistemology the idea of "evidence" has been conceived above all as perceptible and is closely linked to the idea of fact. The concept of "scientific fact" will therefore be analyzed and will reveal how in science, contrary to prevailing opinion, "pure facts" practically do not exist and how the "facts" which scientists talk about in their research are always "facts depending on theory". Subsequently the capacity of "facts" to prove scientific hypothesis will be examined. In the light of more recent epistemological reflection the thesis will be maintained according to which facts are not capable of definitively proving the truth of a theory. Such reflection leads to the conclusion that in medicine "evidence" is always dependent on the theoretical and practical context in which researchers work. In the final part of the paper the epistemological presumptions and ambitions of evidence-based medicine will be examined. This analysis concludes that the epistemology on which evidence-based medicine relies upon does not grasp the true complexity of the scientific methodology and can weaken in doctors the beneficial stimulus that is doubt.