Fault inversion has been documented in many basins worldwide, yet the details of how the initial extensional faults impact on the geometry and growth of the reactivated contractional system is often poorly resolved by the available data. Two-dimensional (2D) and 3D seismic reflection, and well data have been used to chart the evolution of inverted faults from the Taranaki Basin, offshore New Zealand. Sedimentary rocks up to 8 km thick record Late Cretaceous–Paleocene normal faults inverted during Miocene and younger shortening. The displacement and length of early normal faults is a key determinant for the reactivation and size of the subsequent reverse faults. All normal faults with maximum vertical displacements ≥600 m and lengths ≥9 km were inverted along their entire length, while smaller faults were not inverted. The proportion of the total basin-wide strain accommodated on each fault is comparable between deformational episodes. The hierarchy of reverse fault lengths was established rapidly, with longer faults accruing a greater proportion of the total strain from an early stage of shortening. The reverse fault system is dominated by inverted normal faults, which accrue displacement at the expense of smaller faults, and utilize the largest crustal-scale elements of the pre-existing system. The size of pre-existing heterogeneities is an important control for the magnitude and spatial extent of elevated stresses during contraction, which, in turn, control the dimensions, locations and displacements of subsequent fault growth.