ABSTRACT How molecular clouds fragment into dense structures that eventually form stars is an open question. We investigate the relative importance of gravity (both self-gravity and tidal forces) and the volume and surface terms of kinetic, thermal, and magnetic energy for the formation and evolution of molecular clouds and their substructures based on the SILCC-Zoom simulations. These simulations follow the self-consistent formation of cold molecular clouds down to scales of 0.1 pc from the diffuse supernova-driven interstellar medium in a stratified galactic disc. We study the time evolution of seven molecular clouds (of which five are magnetized) over ∼2 Myr. Using a dendrogram, we identify hierarchical three-dimensional substructures inside the clouds with the aim of understanding their dynamics. The virial analysis shows that the dense gas is indeed dominated by the interplay of gravity and turbulence, while magnetic fields and thermal pressure are mostly important for fluffy, atomic structures. However, not all bound structures are gravitationally bound; some are held together by ram pressure aided by other surface terms. Overall, ∼36 per cent of the clouds have >50 per cent of their mass in ‘potentially gravity bound’ structures. A subset of them (70 per cent) is ‘potentially bound’ by gravity on scales >15 pc. A detailed tidal analysis shows that the tidal tensor is highly anisotropic. Yet the tidal forces are generally not strong enough to disrupt either large-scale or dense substructures but cause their deformation. When comparing the tidal and crossing time-scales, we find that tidal forces do not appear to be the main driver of turbulence within the molecular clouds.