Forest cover reduction may affect movements of forest animals, but resistance to animal movements in and out of forests remains unknown despite its importance for modelling. We tested whether ovenbird (Seiurus aurocapilla), a forest-interior songbird, responds similarly to the amount of forest cover while moving locally (~2 km) and over entire landscapes (~25 km). We compared spatially-explicit simulations to field data to address the issue of resistance to movement in open areas. We caught, banded and translocated 143 territorial males 0.8-27 km away from their territory early in the breeding season. Seventy-eight percent and 50% of translocated males returned (homed) within 10 days following "local" and "landscape" translocations respectively. Independent of translocation distance, homing times increased with decreasing forest in the landscape. With a Geographic Information System (GIS), we simulated "least-cost" paths that homing ovenbirds would ideally take, when resistance to movement in open areas ranged 1-1000 times the resistance to movement in forest. The length, the cumulative cost, and variability of simulated least-cost movement paths increased with increasing resistance in open areas. With landscape translocations, least-cost path length explained homing time better than Euclidean distance, and based on an information-theoretic approach, resistance to movement was estimated to be 27 times greater in open areas than in forests (95% confidence interval: 16-45). However, least-cost path length did not perform better than Euclidean distance with local translocations, and the cumulative cost of least-cost paths was not associated to homing time in either translocation scale. We conclude that resistance to animal movements in open areas can be addressed by a combination of GIS modelling and translocation experiments, and is between one and two orders of magnitude greater than resistance to movements in forests, in the case of ovenbirds.