Continental rifting is accommodated by the development of normal fault networks. Fault growth patterns control their related seismic hazards, and the tectonostratigraphic evolution and resource and CO2 storage potential of rifts. Our understanding of fault evolution is largely derived by observing the final geometry and displacement (D)-length (L) characteristics of active and inactive fault arrays, and by subsequently inferring their kinematics. We can rarely determine how these geometric properties change through time, and how the growth of individual fault arrays relate to the temporal evolution of their host networks. Here we use 3D seismic reflection and borehole data from the Exmouth Plateau, NW Shelf, Australia to determine the growth of rift-related, crustal-scale fault arrays and networks over geological timescales (>106 Ma). The excellent-quality seismic data allows us to reconstruct the entire Jurassic-to-Early Cretaceous fault network over a relatively large area (ca. 1,200 km2). We find that fault trace lengths were established early, within the first ca. 7.2 Myr of rifting, and that along-strike migration of throw maxima towards the centre of individual fault arrays occurred after ca. 28.5 Myr of rifting. Faults located in stress shadows become inactive and appear under-displaced relative to adjacent larger faults, onto which strain localises as rifting proceeds. This implies that the scatter frequently observed in D-L plots can simply reflect fault growth and network maturity. We show that by studying complete rift-related normal networks, rather than just individual fault arrays, we can better understand how faults grow and more generally how continental lithosphere deforms as it stretches.