Although birds might increase nesting survival by selecting safe sites, nest predation is a major limiting factor in natural bird populations. We investigated how forest structure, predator distribution, and nest location contribute to nest predation risk, and whether lower predation probabilities characterize formerly "safe" sites. We distributed 304 artificial ground and tree nests in a forest landscape, following the previous year locations of forest grouse observations (with and without broods). We modelled nest predation incidence using generalized mixed models; we also identified potential nest predators by tooth/bill marks on plasticine eggs. We found complex habitat effects: nest predation depended non-linearly on the stand age (low risk in post-clear-cut stands; peak values in middle-aged stands) and there were no simple relationships with predator proximity or nest concealment. Predation probabilities were higher in tree nests than in ground nests, and did not differ in the locations where broods had been observed or not. The results can be partly explained by the behaviour of the most frequent nest predator species, the pine marten, that forages both on the ground and in the canopy and avoids open areas. We suggest that large-scale modifications in forest age structure can alter not only landscape-scale predation pressure on bird nests, but also small-scale distribution and predictability of the predation, which can explain declining numbers of forest-specialist bird species.