Abstract Context By linking species of conservation concern to their abiotic and biotic requirements, habitat suitability models (HSM) can assist targeted conservation measures. Yet, conservation measures may fail if HSM are unable to predict crucial resources. HSM are typically developed using remotely sensed land-cover classification data but not information on resources per se. Objectives While a certain land-cover class may correlate with crucial resources in the area of calibration, political boundaries can abruptly alter these associations. We investigate this potential discrepancy in a well-known study system highly relevant for farmland bird conservation. Methods We compared land cover, land-use intensity and resource availability between plots of highest habitat suitability for little owls (Athene noctua) among two neighbouring, but politically separated areas (i.e. south-western Germany vs. northern Switzerland). Results Land cover and land-use richness did not differ between German and Swiss plots. Yet there were marked differences in terms of land-use intensity and the availability of resources. Land-use intensity was significantly higher and resource availability lower in Swiss compared to German plots. Conclusions While accounting well for remotely sensed data such as land cover, HSM may fail to predict land-use intensity and resources across borders. The relationship between geodata used as proxies and ecologically relevant resources may differ according to history, policies and socio-cultural context, constraining the viability of HSM across political borders. This study emphasises the need for fine-scale resource assessments complementing landscape-scale suitability models. Conservation measures need to consider the availability of crucial resources and their socio-economic moderators to be effective.