Due to the wide variety of substrates it extrudes, P-glycoprotein is one of the main causes of multidrug resistance. Although the chemical structure of a ligand is likely to dictate interactions within the binding site, it is equally important to understand how access to the binding site itself is controlled. Access to the binding site is via an aqueous cavity formed in the main by transmembrane helices (TMHs). These TMHs also have helical structure that extends into the cytoplasm, but the hydrophobic vacuum cleaner model suggests that access to the aqueous cavity is via gates between the membranous regions of TMHs 4 and 6 or between TMHs 10 and 12.Previously, we observed that the intra-membrane sections of these gates tend to close while cytoplasmic regions remain open. We have explored this further via the use of steered MD with substrates initially positioned in the cytoplasmic leaflet of the bilayer. Rather unexpectedly, the simulations show that entrance through the intra-membrane gate requires a significantly larger amount of work compared to the cytoplasmic entrance and with a preference for the cavity between TMH 4 and 6 compared to that between TMH 10 and 12.Once the compound has reached the aqueous cavity, it can freely diffuse towards its binding site. The binding mode was independently assessed through the use of consensus docking and appears to be similar to the so-called “R-site”. The docked conformations were then used to calculate possible unbinding pathways of amitriptyline with the Protein Energy Landscape Exploration (PELE) method.Taken together, these combined approaches suggest that a likely access route to the binding site for many compounds is through a cytoplasmic gate formed by TMHs 4 and 6.