Plant succession theory underpins the development of strategies for the conservation and regeneration of native communities. Current theory has been based largely on space-for-time rather than long-term monitoring data, which have known limitations. There is general consensus that more site-specific studies are needed to corroborate existing hypotheses. The target vegetation is a brigalow (Acacia harpophylla, Mimosaceae) forest in one of Australia's most endangered ecosystems, which was cleared and burnt in 1963. Forty quadrats were placed systematically within each of six 20 m × 20 m permanent plots. Presence, density and per cent canopy cover data were recorded for each species at 18 times over 46 years. Brigalow dominated the original vegetation, assumed dominance soon after clearing through massive root suckering and remained dominant throughout the study. It achieved maximum density within two years when severe intraspecific competition led to self-thinning. After approximately 30 years, vacant niches appeared. Woody understorey species were slow to recolonise. Species richness and other diversity indices increased rapidly to a maximum after 2–4 years, declined until the 30th year when they again increased. This was the pattern of the species-rich herbaceous layer; woody species showed a steady monotonic increase. The ‘hump-shaped’ relationship between cover (biomass) and species richness was confirmed. This example fits the inhibition model for which few examples have been described. While the long-term successional pattern is slightly confounded by climatic variability preceding sample surveys, this space-for-time study not only supports a bimodal pattern of diversity over time but also indicates that the relative species richness of the herbaceous and woody layers may explain the extreme variability reported in the literature.