Introduced grasses are a major threat to dryland ecosystems world-wide because of their ability to transform plant communities and change fire regimes. These structural and functional shifts are often assumed to impact wildlife but this has rarely been measured directly. Likewise, evaluation of weed removal programs rarely considers benefits to fauna, thereby limiting information that could inform management decisions. We used an experimental approach to test the impacts of removing invasive buffel grass (Cenchrus ciliaris), a globally significant invader of dryland systems, on reptiles, a prominent component of the Australian desert fauna. A combination of mechanical and herbicide treatment was applied to replicate plots in areas that had been invaded for at least two decades and changes to ground cover and plant and reptile assemblages were monitored over six years and compared to still-invaded control plots. Following treatment, native plants re-established without the need for reseeding or planting, especially during a period of high rainfall, when positive effects on reptiles also became apparent. The abundance and species richness of reptiles increased at all plots during the mesic period, but less so in control plots, and remained higher at treated plots thereafter, although this was only significant at some times. Post-treatment 27 of 36 species were captured more frequently in treated plots and only four species, all with very low captures, were captured more often in invaded control plots. This consistent trend among species suggests negative impacts of buffel grass on reptiles are likely caused by broad factors such as reduced prey or habitat diversity. Together with concurrent research at the same sites, our results provide experimental evidence that removing buffel grass from heavily invaded areas, even at small scales, benefits a variety of native flora and fauna. Until landscape-scale options are available, restoration of smaller areas within buffel-invaded landscapes can help to preserve native seed banks and adult plants, reduce fire impacts, and provide patches of favourable habitat for fauna. The creation of ‘islands’ of restored native vegetation deserves further consideration as an effective intervention that could help to achieve short and long-term conservation goals in grass-invaded dryland ecosystems.