Reliable information on species distributions and abundances is essential for effective conservation management, yet comprehensive surveys are often constrained by limited resources. Indicators offer a potential solution by inferring population parameters of target species from more easily obtainable data. One promising approach is that of indicators composed of multiple species. While these have shown potential at the community level, their effectiveness for predicting the distribution and abundance of individual species at the scale of protected areas remains poorly understood. We evaluated the performance of multi-species indicators using random forest models for predicting presence-absence and abundance of forest bird species in Domogled-Valea Cernei National Park and tested their spatial transferability to Buila-Vˆ anturarița National Park, both in Romania. We compared multi-species indicator models (multiple species predictors) with single-species indicator models (one species predictor), models based solely on readily available environmental predictors, and combined multi-species indicator-environmental models. Multi-species indicator models consistently outperformed single-species models and environmental-only models. Combining species and environmental predictors improved performance relative to environmental-only models but did not consistently outperform species-only models, suggesting that multi-species indicators captured the most relevant environmental information at this spatial scale. Approximately half of the target species were predicted with at least moderate accuracy in presence-absence models, indicating potential usefulness for reserve-scale management. Abundance models showed weaker performance, although a few species achieved acceptable explanatory power. Spatial transferability was limited: presenceabsence models retained a weak predictive signal in the independent protected area, whereas abundance models showed virtually no transferability. Overall, multi-species indicators can provide useful predictions of species distributions within protected areas, but limited transferability and species-specific variation highlight the need for careful, context-dependent evaluation before application in conservation planning.