Multi-regional electric power systems (MEPS) management involves hierarchical decision-making, multi-dimensional uncertainties, and inter-regional interactions. Conventional deterministic single-level MEPS models struggle to balance trade-offs across multiple objectives and/or criteria at different decision levels and manage uncertainties. To support the low-carbon transition of Canadian MEPS (2031–2050), a factorial non-deterministic multi-stage bi-level programming model (FNDMSBP) is developed, with dual objectives of minimizing GHG emissions at the upper level and system costs at the lower level. By 2050, approximately 80 % of Canada’s electricity will come from non-fossil fuels, with new capacity driven by wind, carbon capture and storage (CCS), and small modular reactor technologies tailored to varying electricity demands. Average annual GHG emissions will decrease to [22.95, 24.87] Mt CO2eq., cutting emissions by over half from 2020 levels. Emission intensity will drop from [0.071, 0.075] to 0.013 kg/kWh, with clean power utilization rising from 0.014 to [0.024, 0.025] MWh/C$. Natural gas and uranium remain essential fuels, contributing over 87 % of energy consumption and 20 % of system costs, with sensitivity analysis highlighting natural gas prices’ significant effects on costs and emissions. Compared to single-level models, FNDMSBP demonstrates its superiority in balancing environmental and economic goals, achieving a GHG reduction of [45.06, 46.87]% with a modest cost increase of [3.56, 3.60]%. This is achieved by prioritizing emission mitigation in high-emission regions such as Alberta and Saskatchewan, while advancing natural gas-CCS and wind energy. This study will help decision-makers analyze interactions and tradeoffs between environmental and economic development under uncertainty, supporting collaborative and sustainable MEPS management.