EN
The paper focuses on one of the problems haunting contemporary thought about possible worlds - namely, that of the relation between conceivability and possibility. 'Conceivability arguments' (based on the premise that the conceivability of a given situation implies its possibility) are often used carelessly, not being properly rooted in any explicit and full-fledged ontology. In the paper the reasons that are commonly proposed for the 'liberal' use of 'conceivability arguments' are critically discussed and certain suggestions as to the proper application of such arguments are presented. It is then pointed out that the issue discussed in fact refers to a deeper problem, that of the opposition between combinatorial and combination ontology. The above is also shown to be a modern version of the classical dispute between Descartes and Leibniz. Finally, the authoress presents some general comments on 'modal epistemology'.