Rehabilitation of post-mining lands frequently aims to create “self-sustaining” systems. Where native vegetation is the designated post-mining land use, it is generally assumed that rehabilitation that is similar to local native ecosystems is more likely to be sustainable. I compared landscape functionality, plant community composition, and vegetation structure in (1) reference sites representing pre-mining native forest; (2) reference sites representing potential landscape analogues for the post-mining landscape; and (3) a 23-year chronosequence of post-mining rehabilitation on the Weipa bauxite plateau, Cape York Peninsula, Australia. The trends across the post-mining chronosequence indicate that vegetation growth is rapid in the first 5–8 years, and then slows with mean height approaching an asymptote after approximately 15 years. Landscape function indices showed a response that coincided with vegetation growth. Vegetation composition was significantly different from reference native forest. Most importantly, from the perspective of creating self-sustaining ecosystems, the contribution of local framework species to vegetation in rehabilitation was significantly lower than in reference native forest. I discuss the results in relation to theoretical models of succession and conclude that without management intervention, differences between post-mining rehabilitation and native forest are likely to be persistent.