Mapping of oil reserves involves the use of seismic lines (linear disturbances) to determine both their location and extent. Conventional clearing techniques for seismic assessment have left a legacy of linear disturbances that cause habitat fragmentation. Little is known, however, about how local and landscape factors affect natural regeneration patterns of trees and shrubs on seismic lines that facilitate mapping and future projections of regeneration patterns. To understand factors affecting early forest regeneration and to predict future trends in regeneration of legacy seismic lines we used LiDAR, forest stand databases and a disturbance inventory of conventional seismic lines to model seismic line regeneration to a 3 m height in a 1806 km2 area in northeastern Alberta, Canada. Regeneration to 3 m was inversely related to terrain wetness, line width, proximity to roads (as a proxy for human use of lines), and the lowland ecosites. Overall, terrain wetness and the presence of fen ecosites had the strongest negative effect on regeneration patterns; the wettest sites failed to recover even after 50 years post-disturbance. Predictions of future regeneration rates on existing lines suggested that approximately one-third of existing linear disturbance footprints in this boreal landscape will remain un-regenerated 50 years later resulting in persistent habitat fragmentation. Model predictions estimating regeneration probability are particularly valuable for estimating current and future forest regeneration trajectories on linear disturbances which are a conservation concern and a focus for restoration and planning by government, industry and conservation organizations.