Although protected areas have been used as principal conservation tools, most of them are suffering from human-induced threats. Consequently, a good understanding of such human-driven threats on biodiversity and identifying early warning systems for habitat change in protected areas is necessary for effective conservation of natural resources. To examine the impact of human disturbance on avifaunal assemblages and to assess the potential application of birds as bioindicators of forest health monitoring in the Afromontane forest of the Bale Mountains of Ethiopia, I recorded birds and habitat variables in three protected and three unprotected forests using a point transect method in 2009 and 2012. The two land use types differ in disturbance levels (higher in the unprotected areas), vegetation structure and bird assemblages. Species richness of entire bird guild, open woodland and open land habitat guilds, granivore and insectivore feeding guilds, and shrub layer and ground layer foraging substrate guilds of birds were significantly higher in the unprotected areas than the protected areas. Abundances of guilds of birds mostly followed a similar trend with species richness. However, densities of overall and forest-specialist bird guilds were higher in the protected area and vice versa for the other guilds. In general, the protected area assemblages were dominated by forest-specialist species, while those of the unprotected areas were dominated by openland and shrubland species. The implication is that disturbance had caused encroachment of non-native species (openland, open woodland and shrub land species) while negatively affecting native species (forest species, particularly tree canopy foragers). These assemblage differences are linked to changes in vegetation structure caused by disturbance. Thus, further forest degradation in the protected area should be avoided in order to maintain native/forest-specialist species. Given the differences in bird assemblages between the two land use types, there is a high likelihood that bioindicator species (i.e. indicator species - those 'characteristic' of a particular habitat - and detector species - those occurring in the different habitats considered but with moderate indication value) can be identified, therefore providing a useful tool to monitor ecosystem health of the forests. Four and nine species were identified as appropriate indicator species (i.e. species with indicator values > 60% and fulfilling biological and niche history criteria used in selection) in the protected and unprotected areas, respectively. In addition, nine species were identified as detectors of habitat change in the protected areas. These bioindicators provide a useful tool for managers of Afromontane forest in the Bale Mountains, as well as similar habitats elsewhere, for long term monitoring of ecosystem health of the forests.