Internal initiation of translation was first demonstrated for the picornaviruses (small positive-stranded RNA viruses, including many pathogens of humans [e.g. poliovirus, hepatitis A virus, human rhinoviruses] and animals [e.g. foot-and-mouth disease virus, encephalomyocarditis virus]). This mode of translation initiation is now known to be operational for other RNA viruses and on an appreciable number of cellular messenger RNAs (mRNAs). It is determined by the IRES (Internal Ribosome Entry Segment): a defined region of RNA within the 5’ untranslated region (5’- UTR), which may extend up to, or even beyond, the authentic initiation codon. Generally IRESes are characterised by being long (about 450 nt. for the picornaviruses), highly structured elements, encumbered by unused initiation codons. However, few sequence or structural similarities between different IRESes can be identified. A general feature of internal initiation of translation is that it is 5’-end independent. Additionally, picornaviral RNAs are naturally uncapped. Thus, the accepted method of recruitement of the 40S ribosomal subunit to the mRNA via a 5’ cap cannot be invoked. Instead, the initial ribosome-RNA interaction occurs within the IRES, and it appears that the location of this interaction is very precise, distinct in each case. This review focuses on how ribosomes are recruited to mRNA, and addresses the question of whether internal initiation of translation is intrinsically different from 5’-end dependent initiation in this respect. Many IRESes are only functional in the presence of proteins other than canonical initiation factors, and much effort has been directed towards the characterisation of functional protein-RNA interactions. Scientists have long dreamt that specific proteins, distinct from the normal translation apparatus, dictate the critical initial ribosome-IRES interaction. Instead, based on our recent data and that from other groups, we propose a model whereby 40S ribosomal subunit recruitment to mRNA is driven by the same complex of eukaryotic initiation factors (eIF4F) for both 5’-end dependent and 5’-end independent translation initiation. The only distinctions are the active form(s) of the complex, and the role played by each participant in the process, both of which vary according to the particular RNA to be translated.