AbstractThis study adapts and applies the evidence‐based approach for causal inference, a medical standard, to the restoration and sustainable management of large‐scale aquatic ecosystems. Despite long‐term investments in restoring aquatic ecosystems, it has proven difficult to adequately synthesize and evaluate program outcomes, and no standard method has been adopted. Complex linkages between restorative actions and ecosystem responses at a landscape scale make evaluations problematic and most programs focus on monitoring and analysis. Herein, we demonstrate a new transdisciplinary approach integrating techniques from evidence‐based medicine, critical thinking, and cumulative effects assessment. Tiered hypotheses about the effects of landscape‐scale restorative actions are identified using an ecosystem conceptual model. The systematic literature review, a health sciences standard since the 1960s, becomes just one of seven lines of evidence assessed collectively, using critical thinking strategies, causal criteria, and cumulative effects categories. As a demonstration, we analyzed data from 166 locations on the Columbia River and estuary representing 12 indicators of habitat and fish response to floodplain restoration actions intended to benefit culturally and economically important, threatened and endangered salmon. Synthesis of the lines of evidence demonstrated that hydrologic reconnection promoted macrodetritis export, prey availability, and juvenile fish access and feeding. Upon evaluation, the evidence was sufficient to infer cross‐boundary, indirect, compounding, and delayed cumulative effects, and suggestive of nonlinear, landscape‐scale, and spatial density effects. Therefore, on the basis of causal inferences regarding food‐web functions, we concluded that the restoration program is having a cumulative beneficial effect on juvenile salmon. The lines of evidence developed are transferable to other ecosystems: modeling of cumulative net ecosystem improvement, physical modeling of ecosystem controlling factors, meta‐analysis of restoration action effectiveness, analysis of data on target species, research on critical ecological uncertainties, evidence‐based review of the literature, and change analysis on the landscape setting. As with medicine, the science of ecological restoration needs scientific approaches to management decisions, particularly because the consequences affect species extinctions and the availability of ecosystem services. This evidence‐based approach will enable restoration in complex coastal, riverine, and tidal‐fluvial ecosystems like the lower Columbia River to be evaluated when data have accumulated without sufficient synthesis.