In this paper, three main issues are discussed: (i) How the Semantics- Pragmatics Interface (SPI) is supposed to work? (ii) Why do we need a SPI? (iii) Where is the SPI located? I show that the S-P border is porous, and that some inferred meanings are more semantic than pragmatic and vice versa. Secondly, the SPI has as a main function to allow quick and efficient information transfer, from non-linguistic source to linguistic one, and vice and versa. Finally, the SPI is mainly a linguistic issue: semantic meaning is the locus of pragmatic processes, which implies that its conceptual vs. procedural nature has some impacts on the way pragmatic meaning derivations are obtained.