The paper attempts to give an account of the fallibility of medical judgments without recourse to the incompleteness of scientific knowledge. It is argued that because of the inexactness of observations and thus the existence of borderline cases any theory applied for explanation and predicition will produce some false results. This state of affairs is independent of the nature of a theory, i.e., it applies both for non-probabilistic and for probabilistic theories. Some epistemological issues and consequences with regard to a better understanding of clinical reasoning are discussed and the view is compared with the theory of medical fallibility of Gorovitz and MacIntyre.1