We address effects of large-scale forestry on landscape structure and the structure and composition of boreal bird communities in North Sweden. Specifically, we ask: after controlling for the effect of patch size, forest age and tree species composition, is there any residual effect attributable to the reduction in area of old forest? Pairs of landscape blocks (25 by 25 km) were selected to maximize area difference in human-induced disturbance, clear-cut as opposed to semi-natural old forest. Median distance to natural edge (wetlands, open water) from randomly selected points in forest was 250 and 200 m in high and low impact landscapes, respectively, indicating a high degree of 'natural' fragmentation of the pristine boreal landscape in the area. By contrast, median distance to clear-cut in uncut forest was 750 and 100 m, respectively. Clear-cuts in high impact landscapes were disproportionally more common in areas with contiguous forest land than in areas with spatially disjunct forest, implicating that forestry increases natural fragmentation of the landscape by subdividing larger forest tracts. Point counts along forestry roads showed that species richness and relative abundance of forest birds were higher in landscapes with low forestry impact. These differences can partly be explained by differences in age composition of forest and composition of tree species. After controlling for patch size, forest age and tree species composition, a significant effect of forestry impact remained for Sibirian species and the Tree pipit Anthus trivialis. Our results thus imply that this group of species and the Tree pipit may be sensitive to forest fragmentation. In contrast to previous Finnish studies, we found relatively small negative effects on relative abundance of species hypothesized to be negatively affected by large-scale clear-cutting forestry. However, our picture of the present does not contradict results from Finnish long-term population studies. Five factors may account for this: 1) clear-cut areas are not permanently transformed into other land use types, 2) planted forests are not completely inhabitable for species preferring older forest, 3) the majority of species in the regional pool are habitat generalists, 4) the region studied is still extensively covered with semi-natural forest, and 5) our study area is relatively close to contiguous boreal forest in Russia, a potential source area for taiga species.