The impact of a pre-existing rift fabric on normal fault array evolution during a subsequent phase of lithospheric extension is investigated using 2-D and 3-D seismic reflection, and borehole data from the northern Horda Platform, Norwegian North Sea. Two fault populations are developed: (i) a population comprising relatively tall (>2 km), N-S-striking faults, which have >1.5 km of throw. These faults are up to 60 km long, penetrate down into crystalline basement and bound the eastern margins of 6–15 km wide half-graben, which contain >3 km of pre-Jurassic, likely Permo–Triassic, but possibly Devonian syn-rift strata; and (ii) a population comprising vertically restricted (<1 km), NW-SE-striking faults, which are more closely spaced (0.5–5 km), have lower displacements (30–100 m) and not as long (2–10 km) as those in the N–S-striking population. The NW-SE-striking population typically occurs between the N-S-striking population, and may terminate against or cross-cut the larger structures. NW–SE-striking faults do not bound pre-Jurassic half-graben and are largely restricted to the Jurassic-to-Cretaceous succession. Seismic-stratigraphic observations, and the stratigraphic position of the fault tips in both fault populations, allow us to reconstruct the Late Jurassic-to-Early Cretaceous growth history of the northern Horda Platform fault array. We suggest the large, N-S-striking population was active during the Permo–Triassic and possibly earlier (Devonian?), before becoming inactive and buried during the Early and Middle Jurassic. After a period of relative tectonic quiescence, the N-S-striking, pre-Jurassic fault population propagated through the Early-Middle Jurassic cover and individual fault systems rapidly (within <10 Ma) established their maximum length in response to Late Jurassic extension. These fault systems became the dominant structures in the newly formed fault array and defined the locations of the main, Late Jurassic-to-Early Cretaceous, syn-rift depocentres. Late Jurassic extension was also accommodated by broadly synchronous growth of the NW-SE-striking fault population; the eventual death of this population occurred in response to the localization of strain onto the N–S-striking fault population. Our study demonstrates that the inheritance of a pre-existing rift fabric can influence the geometry and growth of individual fault systems and the fault array as a whole. On the basis of observations made in this study, we present a conceptual model that highlights the influence of a pre-existing rift fabric on fault array evolution in polyphase rifts.