Long term analyses on habitat dynamics are challenging when data from the past is scarce. In addition, large scale perspectives are difficult to attain without proper modelling of habitat conditions. These drawbacks were overcome in the case of capercaillie in Romania with the help of satellite-based information and by finding relevant environmental variables which influence the species habitat. We used MaxEnt software with four variables (spruce distribution, elevation, precipitation and temperature in June) and 750 presence points. The model had a high predictive ability (area under the curve over 95%), elevation and spruce distribution being the most important variables. A separate set of 902 presence points was used for model crosschecking. Most of the crosscheck points (98.6%) fell inside the predicted habitat confirming the model performance. To detect habitat preferences the vegetation mosaic in 90 circles (3 km-radius) was analyzed using both historical (Cold War) and recent satellite information. The modelled habitat across the Romanian Carpathians was a mosaic of stands of different ages. Moreover, a similar structure was found inside and outside the analyzed circles: forests between 65 and 120 yr. dominated, followed by the younger classes (20–65 yr. and forests <20 yr.). The oldest class (> 120 yr.) was the least common. The majority (i.e., 90%) of the presence points used for model crosscheck fell within actively managed forests, including former large clearcuts (revealed by the Cold War spy-satellite images). This suggests that the species was tolerant to forest management and also able to recolonize areas when habitat became favorable again. Therefore, controlled forest management maintains the habitat favorability at very large scales in a shifting steady-state mosaic. This management-driven habitat together with the one produced by natural disturbances in natural forests provides favorable conditions for a habitat continuum along the Romanian Carpathians.