The composition of mammalian communities in Australia's Eucalyptus forests and woodlands is known to be affected by fire. However, there are few published studies that compare mammal assemblages in recently burnt and long-unburnt forests because there are few areas with long-term fire history data. Understanding the value of long-unburnt forest is important because it is becoming rare in fire-prone regions of the world, such as south-eastern Australia, partly because of the widespread use of prescribed burning. We deployed wildlife cameras for 28 trap-nights at each of 81 sites that ranged from 0.5 to at least 96 years since the last fire. We recorded a total of 15 native mammal species. At least one mammal species was recorded at 80 of the 81 sites. Significantly more species were detected at long-unburnt sites (>96 years since fire) than sites 0.5–12 years since the last fire. Species composition varied significantly between sites 0.5–12 years and >96 years since the last fire but did not vary between sites 0.5–2 years and 6–12 years since the last fire. Although there was not one category of time since fire (i.e. 0.5–2 years, 6–12 years and >96 years) in which all 15 native mammal species were recorded, long-unburnt sites were significantly more important for the occurrence of seven mammal species; intermediate and recently burnt sites were significantly more important for one species. Our results suggested that, while a diversity of fire ages is important for conserving mammalian diversity, long-unburnt forests and woodlands (which comprised only 8% of our study area) are disproportionately important for mammal conservation. Our results add to a growing body of the literature from south-eastern Australia, suggesting that remaining long-unburnt forest should be afforded protection from fire and more forest should be transitioned to long unburnt.