We studied the change in carabid species composition across a stand age chronosequence in exotic Pinus radiata plantations in the central North Island of New Zealand. From 2002 to 2004, we pitfall-trapped beetles along 580-m transects that traversed replicate stands of each of 1, 2, 4, 8, 16 and 26+ years old and their adjacent 26+-year-old reference stand. We captured a total of 13,101 individuals of 19 species, and there was significant variation in activity density over the chronosequence, with activity density peaking 2 years after harvest and then declining consistently with increasing stand age. However, after correcting for variation in sample abundance, no differences in species richness of native or exotic carabids were detected across the age sequence. Nevertheless, carabid assemblages in young regenerating (1, 2, and 4-year) stands were significantly different in composition from adjacent older stands (largely due to colonisation by open habitat species), and from each other. The spatial change in carabid composition in young stands relative to adjacent mature forest was rapid and typically occurred within a distance of just 5m of the stand boundary. The carabid fauna of older stands was dominated by native forest generalist species, with few species indicative of stands of a particular age. Cicindela tuberculata was the only species to show a strong affinity for a particular age class (in this case 2-year-old-stands), however other open habitat species were closely associated with 1 and 2-year-old-stands. To conserve the native biodiversity remaining in P. radiata plantations and in adjacent natural forest remnants we recommend maintaining connectivity among stands of at least 8 years of age, or older, that essentially contain a generalist native forest carabid fauna that is distinct from the disturbance-adapted fauna in young stands. However, extrapolation beyond our carabid data will require additional validation of these results for other taxa. To expedite this process we suggest a focus on disturbance-sensitive species (e.g., those most sensitive to forest harvesting) that can be monitored to define recovery times of species post-harvest.