Can the 'continuity thesis' upheld by R Boudon - that rationality is an undivided and permanent property of human thought in all fields in which it operates - include religious beliefs, which are often viewed as those most unrelated to any rationality ? The author will begin by distinguishing two types of reason for endorsing a belief, that is for holding a statement to be true or likely. The first type concerns the processes by which a belief is acquired, and are therefore exterior to it ; the second are independent of the acquisition process, and concern cognitive processes used to test the internal contents of this belief. In the latter case, probably the more characteristic of our societies, the author will examine the rationality of the arguments most often used by believers themselves in order to justify their adherence to beliefs : the cognitive argument of the evidence for them and the instrumental argument of their practical efficacy. Then the author will try to determine to what extent these arguments are consistent with a rational conception of belief behaviours (is it possible to believe something for its usefulness ?) and if, as Boudon's conception leads us to suppose, they may or may not be reduced to rationality of the 'cognitive' type. Finally, the author will examine how well the 'cognitive' hypothesis stands up when one takes into account the specific semantics of religious beliefs, that is to say, the 'semi-propositional' character of most of their statements.< Réduire