The paper deals with such a modification of genuine modal realism as to accommodate impossible worlds into its ontology. First of all, the theory of modal realism is presented. Next, several motivations for the acceptance of impossible worlds are adduced. In particular, I point out that genuine modal realism suffers from the so-called granularity problem – distinct impossible as well as necessarily coextensive properties and propositions are (unintuitively) identified – and in order to avoid it, the world semantics should be extended by impossible worlds. Furthermore, Lewis’s argument against impossible worlds is presented. It is argued that the argument can be weakened by rejection one of its premises. Finally, I counter two objections against the proposal. Although my strategy accounts for the Opinion concerning the impossible, it allegedly violates another Opinion that reality is classical. But, as it seems, there is no no-question-begging reason to think that reality is classical. How can we know, after all, which logic appropriately describes reality? Without a definite answer to the question, the objection from incredibility then simply collapses into a statement of possibilist dogma.