Galaxy clusters are the most massive objects in the cosmic web and their distribution and properties are a sensitive cosmological probe. Hydrodynamic simulations have now evolved to point where the modelling of subgrid processes, such as star formation, supernovae and AGN feedback, yield clusters that reproduce the wide range of observed scaling relations, including their intrinsic scatter. However, due to the finite size of the simulations the rarest and most massive clusters are absent from these results. We introduce the MACSIS project, which uses a recalibrated cosmo-OWLS model (McCarthy et al. in prep.) to perform ‘zoomed’ simulations of the most massive clusters in the Universe. We examine how the scaling relations extend to the most massive objects and compare our results with the observations. We also investigate the evolution of the scaling relations with redshift as, due to their rarity, these systems are most likely to be observed at high redshift. Finally, we perform a morphological analysis to select a subset of dynamically relaxed clusters and examine the bias it introduces to the scaling relations.