Large-scale urbanisation has become a significant barrier to the natural migration of tree species, which is being exacerbated by accelerated climate change. Within this context, improving the permeability of urban landscapes is expected to be an effective strategy to facilitate the process of forest migration through cities. This study develops a method to assess the permeability of urban green spaces as stepping stones for forest migration, from the perspective of seed dispersal. The proposed method combines a least-cost path model and a graph theory-based approach. The least-cost path model is applied to map the potential pathways of seed dispersal at multiple spatial and temporal scales, based on which graph theory-based indices are used to quantify the accessibility of urban landscapes for seed dispersal agents. This method is demonstrated by a case study in the Greater Manchester area, UK. Eurasian jay, Eurasian siskin, coal tit and grey squirrel are selected as the main seed dispersal agents in the study area. The results provide a comparison of the landscape permeability maps generated from different seed dispersal agents and identify key areas likely to facilitate the process of forest migration. Recommendations regarding landscape management for improving permeability are also discussed.