Despite methodological and theoretical advances in conservation genetics, data on genetic variation on broad regional spatial scales are still scarce, leading conservation planners to use general heuristic or simulation models for an integrated analysis of genetic, demographic and landscape parameters. Here, we extended previous results by evaluating spatial patterns of extinction by inbreeding depression under stochastic variation of environments for mammalian populations in 31 conservation units of the Brazilian Cerrado. We observed a large spatial variation of times to extinction, for different conservation units and body-size classes of species. For small-bodied species (500 g), the population times to extinction in the conservation units were usually longer than 200 years, whereas for medium-bodied (5 kg) and large-bodied (50 kg) species this time was considerably shorter, and only a few units would maintain viable populations for more than 100 years. These figures are consistent with the current status of mammalian conservation in Brazil, and hopefully the simulated scenarios can be integrated to patterns of human occupation and habitat loss in the biome, in order to furnish overall guidelines for biodiversity conservation.Key words: population persistence, population viability analysis, mammals, Cerrado, conservation units. Evaluating broad-scale patterns in genetic diversity was the background for the development of "landscape genetics", a new discipline in Genetics that aims to provide information on how landscape and environmental features influence population genetic structure (Manel et al., 2003;Pearse and Crandall, 2004;Kidd and Ritchie, 2006;Storfer et al., 2006). The basic idea is to evaluate spatial patterns of genetic variation and to test for correlations between genetic discontinuities among local populations and landscape-level features, especially those related to recent human occupation (Manel et al., 2003), in order to use this knowledge for developing better conservation strategies (e.g., Telles, 2002, 2006). Its implications on conservation are based on the assumption that habitat fragmentation, loss or conversion, generated by different types of human activities, can lead local populations to extinction, because of the decreasing levels of gene flow and reduction in effective population sizes (Fahring, 2003). On the other hand, despite methodological and theoretical advances provided by landscape genetics, data on genetic variation on broad regional spatial scales are still scarce, and therefore conservation planners can use overall and heuristic models which allow an integrated analysis of genetic, demographic and landscape parameters that, in turn, can provide data for an evaluation of biodiversity persistence and extinction risks (Henle et al., 2004). These models are usually parameterized by real-life history parameters of the target species or populations and by the environmental characteristics of the region in which the populations are conserved (Frankham et al., 2002;Mo...