Restoring tropical rainforests is becoming increasingly urgent. However, in most restoration plantings it is not possible to include the full suite of species found in the original rainforest. Full recovery of species composition thus depends on the dispersal and recruitment of species that are not planted. In many restoration projects, however, recruitment is dominated by a low diversity of regionally-abundant pioneer species and species with small, easily dispersed seeds. These species are characteristic of secondary rainforest and do not include the far more diverse suite of species characteristic of the original, primary rainforest. Such primary rainforest species are usually more vulnerable to the effects of fragmentation than disturbance-adapted pioneers and thus are of greater conservation concern, as well as being required for the full recovery of many important ecosystem functions. As restoring ecosystem processes is one of the central goals of restoration, this raises the question of which, if any, of the available rainforest restoration methods may be used to promote the recruitment of primary rainforest species. We compared the species composition and functional group composition of recruited trees and shrubs in a 25-year-old restoration experiment with those of the originally planted trees, and with nearby primary rainforest and secondary rainforest reference sites in an area of upland rainforest in north-eastern Australia. Our objective was to compare the performance of four commonly-used restoration methods: (i) unassisted (passive) regeneration, (ii) Pioneer Monoculture plantings, (iii) Framework Method plantings, and (iv) Maximum Diversity plantings. The species composition and functional group composition of recruited individuals within all treatments were similar to those of secondary rainforest and highly dissimilar to both primary rainforest and plantations. Pioneer species, species with small, biotically-dispersed diaspores, and canopy trees were over-represented among both recruited individuals and in secondary rainforest. Conversely, climax species, species with large, biotically-dispersed diaspores, species with abiotically-dispersed diaspores, and understorey trees were under-represented among both recruited individuals and secondary rainforest. Restoration treatments had little effect on the species or functional group composition of recruited individuals. Our results indicate that species from nearby primary rainforest almost completely failed to recruit into any of the restoration treatments. We argue that this failure was most likely due to the absence of frugivores able to disperse larger diaspores from both secondary forest and restored forest in our study site. Further direct management intervention will be required to restore primary rainforest plant species to restored forests in this region.