The pyrodiversity–biodiversity (P–B) hypothesis posits that spatiotemporally variable fire regimes increase wildlife habitat diversity, and that the fine-grained mosaics resulting from small patchy fires enhance biodiversity. This logic underpins the patch mosaic burning (PMB) paradigm and reinforces the benefits of Indigenous fire management, which tends to promote pyrodiversity. However, tests of the P–B hypothesis and PMB paradigm are few. One of the most comprehensive field evaluations—a snapshot study of pre-existing fire mosaics in south-east Australian semi-arid mallee eucalypt woodlands—found little support. To explore the longer-term effects of fire mosaic grain size on habitat availability and biodiversity, we combined published data from the mallee study with a simple fire simulation. We simulated 500 years of landscape burning under different fire sizes. In the resulting mosaics, we assessed the proportional mixture and patch configuration of successional habitat states, then summarised habitat availability through time using a composite index based on the published fire history responses of 22 vertebrate taxa from the mallee study. Small fires formed fine-grained mosaics with a stable habitat mixture and with habitat diversity occurring at fine scales. Large fires formed coarse-grained mosaics with the opposite properties. The fine-grained mosaics maintained optimal habitat availability for vertebrate diversity over 500 years, while the fluctuating habitat mixture in the coarse-grained mosaics was unlikely to maintain maximum vertebrate diversity. Broadly, our results support the P–B hypothesis and justify further field-testing and evaluation of PMB programs to manage both pyrodiversity and biodiversity in the mallee and other flammable landscapes.