G protein-coupled receptors (GPCRs) are proteins of pharmaceutical importance, with over 30% of all drugs in clinical use targeting them. Increasing numbers of X-ray crystal (XRC) structures of GPCRs offer a wealth of data relating to ligand binding. For the β-adrenoceptors (β-ARs), XRC structures are available for human β2- and turkey β1-subtypes, in complexes with a range of ligands. While these structures provide insight into the origins of ligand structure-activity relationships (SARs), questions remain. The ligands in all published complexed XRC structures lack extensive substitution, with no obvious way the ligand-binding site can accommodate β1-AR-selective antagonists with extended side-chains para- to the common aryloxypropanolamine pharmacophore. Using standard computational docking tools with such ligands generally returns poses that fail to explain known SARs. Application of our Active Site Pressurisation modelling method to β-AR XRC structures and homology models, however, reveals a dynamic area in the ligand-binding pocket that, through minor changes in amino acid side chain orientations, opens a fissure between transmembrane helices H4 and H5, exposing intra-membrane space. This fissure, which we term the “keyhole”, is ideally located to accommodate extended moieties present in many high-affinity β1-AR-selective ligands, allowing the rest of the ligand structure to adopt a canonical pose in the orthosteric binding site. We propose the keyhole may be a feature of both β1- and β2-ARs, but that subtle structural differences exist between the two, contributing to subtype-selectivity. This has consequences for the rational design of future generations of subtype-selective ligands for these therapeutically important targets.