Evaluating alternative future scenarios using simulation models is an emerging approach to conservation planning over large spatial and temporal extents. Such an approach is useful when predictions cannot be validated empirically; however, evaluating the sensitivity of scenario-based approaches to key uncertainties is necessary so that managers understand how real-world constraints might impact results. We used a simulation approach to investigate the sensitivity of freshwater habitat restoration plans to spatial pattern (e.g., project locations) and economic uncertainties (e.g., budget level and project costs) in a case study on Pacific salmon in the Lewis River Basin (WA, U.S.A.). We found that some evaluation metrics performed better when restoration was distributed randomly, whereas others improved more when restoration occurred on contiguous stretches of river. Subwatersheds responded differently to restoration, suggesting that intrinsic characteristics limit restoration potential. Changes in budget usually caused proportionate improvements in habitat but disproportionate fish responses, suggesting a cumulative benefit for fish. Most per-dollar benefits increased as project cost decreased; therefore, model predictions are sensitive to project cost. Simulation analyses provided increased information about how economics and spatial pattern might affect habitats and salmon; these insights can help policymakers identify an appropriate restoration strategy.