The knowledge on the sorption behaviour of antibiotics on nanomaterials is limited, especially regarding the reaction mechanism on the surface of carbon nanomaterials, which may determine both the adsorptive capacity and regeneration efficiency of graphene adsorbers. In this work, we used molecular modelling to generate the most comprehensive (to date) adsorption dataset for pristine and functionalised graphene interacting with 8 β-lactams, 3 macrolide, 12 quinolone, 4 tetracycline, 15 sulphonamide, trimethoprim, 2 lincosamide, 2 phenicole and 4 nitroimidazole antibiotics, and their transformation products in water and n-octanol. Results show that various non-covalent interactions that operate simultaneously, including van der Waals dispersion forces, π-interactions, hydrophobic interaction and hydrogen bonding, facilitate adsorption. The molecular properties of antibiotics and graphene/graphene oxide, as well as the composition of the background solution regulate the magnitude of these interactions. Our findings demonstrate that the most efficient method for the removal of antibiotics from aquatic environments is the use of graphene at environmental pH. The subsequent regeneration of the sorbent is best achieved through washing with slightly basic (pH 8–10) non-polar solvents. The obtained theoretical insights expand and complement experimental observations and provide important information that can contribute to further exploration into the adsorbent properties of graphene-based materials, and towards the development of predictive adsorption models.