The aim is to motivate theoretically a relevance approach to (indicative) conditionals in a comparative discussion of the main alternatives. In particular, it will be argued that a relevance approach to conditionals is better motivated than the suppositional theory currently enjoying wide endorsement. In the course of this discussion, an argument will be presented for why failures of the epistemic relevance of the antecedent for the consequent should be counted as genuine semantic defects (as opposed to be relegated to pragmatics). Furthermore, strategies for dealing with compositionality and the perceived objective purport of indicative conditionals will be put forward.