Several control mechanisms of eukaryotic gene expression target the initiation step of mRNA translation. The canonical translation initiation pathway begins with cap-dependent attachment of the small ribosomal subunit (SSU) to the messenger ribonucleic acid (mRNA) followed by an energy-dependent, sequential ‘scanning’ of the 5′ untranslated regions (UTRs). Scanning through the 5′UTR requires the adenosine triphosphate (ATP)-dependent RNA helicase eukaryotic initiation factor (eIF) 4A and its efficiency contributes to the specific rate of protein synthesis. Thus, understanding the molecular details of the scanning mechanism remains a priority task for the field. Here, we studied the effects of inhibiting ATP-dependent translation and eIF4A in cell-free translation and reconstituted initiation reactions programmed with capped mRNAs featuring different 5′UTRs. An aptamer that blocks eIF4A in an inactive state away from mRNA inhibited translation of capped mRNA with the moderately structured β-globin sequences in the 5′UTR but not that of an mRNA with a poly(A) sequence as the 5′UTR. By contrast, the nonhydrolysable ATP analogue β,γ-imidoadenosine 5′-triphosphate (AMP-PNP) inhibited translation irrespective of the 5′UTR sequence, suggesting that complexes that contain ATP-binding proteins in their ATP-bound form can obstruct and/or actively block progression of ribosome recruitment and/or scanning on mRNA. Further, using primer extension inhibition to locate SSUs on mRNA (‘toeprinting’), we identify an SSU complex which inhibits primer extension approximately eight nucleotides upstream from the usual toeprinting stop generated by SSUs positioned over the start codon. This ‘−8 nt toeprint’ was seen with mRNA 5′UTRs of different length, sequence and structure potential. Importantly, the ‘−8 nt toeprint’ was strongly stimulated by the presence of the cap on the mRNA, as well as the presence of eIFs 4F, 4A/4B and ATP, implying active scanning. We assembled cell-free translation reactions with capped mRNA featuring an extended 5′UTR and used cycloheximide to arrest elongating ribosomes at the start codon. Impeding scanning through the 5′UTR in this system with elevated magnesium and AMP-PNP (similar to the toeprinting conditions), we visualised assemblies consisting of several SSUs together with one full ribosome by electron microscopy, suggesting direct detection of scanning intermediates. Collectively, our data provide additional biochemical, molecular and physical evidence to underpin the scanning model of translation initiation in eukaryotes.