Biodiversity is in severe decline globally, attributed in a large part to anthropogenic land use change. The conservation literature refers to the landscape scale as important in mediating biodiversity. However, environmental impact assessment (EIA), a prevalent tool to inform decision-making with respect to ecological considerations such as biodiversity impacts, rarely takes a landscape perspective. Decisions are often made for individual projects, at local scales, with little attention paid to landscape contexts. The cumulative impact of project-by-project decision-making all too often results in alteration of ecological networks in the landscape with associated losses in biodiversity. A disconnect is apparent between scales of analysis for biodiversity conservation and those used for impact assessment. Landscape ecology studies landscape patterns and processes at a range of scales and has potential to bridge this disconnect. This thesis examines the potential to improve biodiversity conservation by better incorporating landscape ecology-based analysis into project EIA. The mixed-methods research follows three lines: (1) identifying gaps between the science of landscape ecology and the practice of EIA, (2) examining the challenges faced by EIA practitioners when considering broader-scale analysis in EIA and associated opportunities for overcoming them, and (3) testing an accessible approach to landscape analysis that incorporates a scenario-based simulation model of cumulative project decision-making. Research was focused on Ontario, Canada, and its multi-jurisdictional EIA regime. Results revealed gaps in how landscape context was considered in EIA, such as the ability of the whole landscape to support species movement and dispersal, and in comparing project-induced land use change to landscape-based ecological targets and thresholds. Quantitative and spatial analyses were infrequently used to assess landscape composition and configuration. Challenges exacerbating these gaps are both policy- and science-based. Weak policy and guidance for broader-scale analysis and a lack of multi-level policy support undermine practitioners' ability to incorporate landscape analysis into EIA. Better multi-jurisdictional data and data management systems are recommended, as well as increasing knowledge of ecological thresholds within the science-practitioner communities. If these challenges can be overcome, the modelling exercise demonstrated that incorporating even simple landscape considerations in project-based decision-making can have a positive effect on biodiversity indicators.