Abstract This opportunistic study compares the vegetation, fuel loads and vertebrate fauna of part of a 120-ha block of tropical open forest protected from fire for 23 years, and an adjacent block burnt annually over this period. Total fuel loads did not differ significantly between the unburnt and annually burnt sites, but their composition was markedly different, with far less grassy fuel, but far more litter fuel, in the unburnt block. There were major differences between treatments in the composition of trees and shrubs, manifest particularly in the number of stems. There was no overall difference in plant species richness between the two treatments, but richness of woody species was far higher in the unburnt treatment, and of annual and perennial grasses, and perennial herbs in the annually burnt treatment. Change in plant species composition from annually burnt to unburnt treatment was directional, in that there was a far higher representation of rainforest-associated species (with the percentage of woody stems attributable to ‘rainforest’ species increasing from 24% of all species in the annually burnt treatment to 43% in the unburnt treatment, that of basal area from 9% to 30%, that of species richness from 8% to 17%, and that of cover from 12 to 47%). The vertebrate species composition varied significantly between treatments, but there was relatively little difference in species richness (other than for a slightly richer reptile fauna in the unburnt treatment). Again, there was a tendency for species that were more common in the unburnt treatment to be rainforest-associated species. The results from this study suggest that there is a sizeable and distinct set of species that are associated with relatively long-unburnt environments, and hence that are strongly disadvantaged under contemporary fire regimes. We suggest that such species need to be better accommodated by fire management through strategic reductions in the frequency of burning.