AbstractAim We used abiotic environmental variables and historical locality records to infer distributions of endangered anuran species of Costa Rica to promote efficient strategies for future amphibian surveys.Location Costa Rica.Methods We used a Maximum Entropy Algorithm (Maxent) to predict potential distribution maps for 17 species of endangered anurans and create a consensus map of species richness. We compared the environmental conditions from localities where relictual amphibian populations were recently rediscovered with the conditions across their historical range to evaluate the possibility that these relictual populations might occur in specific climatic conditions that could explain their persistence. We used a multicriteria analysis considering the following factors: the intersection zones between the consensus map, conservation areas, potential Batrachochytrium dendrobatidis (Bd) distribution, collecting effort and areas within the precipitation range at which reappearances had occurred to locate sites for future surveys.Results The resulting predictions suggest that suitable areas for the highest number of species occur between 1300 and 2500 m.a.s.l and are concentrated along the Pacific slopes of the Cordillera de Talamanca and Cordillera Volcánica Central. Around 45% of the high potential richness area is under protection. Relictual populations of declined species seem to persist mainly in highly humid localities (2500–3500 mm of mean annual precipitation). Around 240 km2 has an ideal environment for the rediscovery of relictual populations. The multicriteria analysis showed that around 0.5% of the Costa Rican territory should be surveyed exhaustively for frogs.Main conclusions Many of the potential refugia we identified here have not been surveyed since 2000, the areas identified by the best model predictions correspond well with the localities of the relictual populations recently reported. We suggest future surveys of missing amphibian species should focus on these areas. The discovery of populations of endangered species can be used to propose conservation areas.