Introduced browsing animals negatively impact New Zealand’s indigenous ecosystems. Eradicating introduced browsers is currently unfeasible at large scales, but culling since the 1960s has successfully reduced populations to a fraction of their earlier sizes. Here we ask whether culling of ungulates has allowed populations of woody plant species to recover across New Zealand forests. Using 73 pairs of permanent fenced exclosure and unfenced control plots, we found rapid increases in sapling densities within exclosures located in disturbed forests, particularly if a seedling bank was already present. Recovery was slower in thinning stands and hampered by dense fern cover. We inferred ungulate diet preference from species recovery rates inside exclosures to test whether culling increased abundance of preferred species across a national network of 574 unfenced permanent forest plots. Across this network, saplings were observed irrespective of their preference to ungulates in the 1970s, but preferred species were rarer within disturbed sites in the 1990s after long-term culling and despite nationwide increases in sapling densities. This indicates that preferred species are relatively heavily affected by browsing after culling, presumably because remaining animals will increase consumption of preferred species as competition is reduced. Our results clearly suggest that culling will not return preferred plants to the landscape immediately, even given suitable conditions for regeneration. Complete removal of ungulates rather than simply reducing their densities may be required for recovery in heavily browsed temperate forests, but since this is only feasible at small spatial scales, management efforts must target sites of high conservation value.