The future of forest biodiversity is increasingly dependent on the ability of species to persist in isolated forest remnants embedded within a mosaic of human land uses. We review the impacts of forest loss, habitat degradation and alteration of the spatial structure of the landscape on connectivity and population persistence in fragmented landscapes. Conservation management is just beginning to come to grips with the challenges faced in moving away from the traditional patch-focused approach of conservation in gazetted nature reserves towards a landscape-focused approach of conserving biodiversity in a mosaic that includes managed and semi-natural habitats. Contentious debate has centred on the ability of degraded forest systems to sustainably support forest specialist species, and the utility of traditional species–area approaches to predict biodiversity loss relationships without appropriately accounting for varying species responses to landscape matrix quality and edge effects. While there is no doubt that some area-sensitive and disturbance-sensitive species require large areas of primary forest to ensure population persistence, there is an emerging consensus that the maintenance of biodiversity will depend as much, if not more, on the extent, magnitude and spatial structuring of landscape processes within the degraded matrix surrounding primary forest remnants. Accordingly, biodiversity conservation must refocus more on the interaction between patch and landscape processes than on patch processes per se. This will demand more effective discrimination of the relative importance of total habitat loss, declining habitat quality and altered spatial structuring of suitable habitat as mechanistic drivers of biodiversity loss in fragmented forest landscapes.