Five main conclusions arise from this review of the responses of species to anthropogenic disturbance in Madagascar: First, species' reactions to anthropogenic disturbance are generally negative, but remain poorly known. Our knowledge is patchy among and within higher taxonomic groups; we are still largely gathering case studies. Second, taxonomic groups vary considerably in which proximate factors are most important. Third, several groups show differing responses within different ecoregions. Whether these differences are consistent across groups requires further testing. Fourth, related species often have divergent reactions to disturbance, even within lower taxonomic groupings (families or genera). Thus, we cannot rely on phylogenetic relatedness or even ecological similarity to infer similarity in responses. Finally, disturbance typically reduces species diversity (especially of native and/or endemic species), but also causes species turnover, typically with forest specialists replaced by grassland generalists, and endemics replaced by non-endemics (including invasives). Given these knowledge gaps, we stress the urgency of applied studies that assess species' ecology, behaviour and health across disturbance gradients, including purely anthropogenic landscapes. Remaining natural vegetation and protected areas will be unable to preserve Madagascar's biodiversity under the impact of climatic change; we must understand responses of plants and animals to disturbance in order to create buffer zones and corridors combining secondary, degraded and natural habitats.