Climate change is a global stressor that can undermine water management policies developed under the assumption of stationary climate, necessitating robust solutions to reducing the risk of system failures for uncertain future climates. While the response-surface-based assessments have provided convenience to explore responsive behaviours of expected system performance to climatic stresses, they were unable to predict the risk of system failures from individual climate projections. In this study, we proposed to use the logistic regressions for evaluating the probability of non-successive outcomes against pre-defined thresholds directly from climate projections, which may be more informative for decision making processes than the expected performances. As a case study, water supply and ecological reliabilities within a large river basin were assessed by combining the eco-engineering decision scaling framework and the logistic regressions. The impact assessment for the South Korean river basin showed that optimal water supply performance at the sub-basins were expected to be satisfactory for the upcoming 20 years of 2020–2039, while the human-demand-only operations could lower the ecological reliabilities. When considering ecological demands in water operations to reduce the ecological vulnerabilities, the stakeholders should accept increasing risks of unsatisfactory supply at the sub-basins with low demands. This study highlights that binary conversions of the performance metrics from the stress tests allow users to measure the risks of system failures varying across sub-components and standpoints with minimal computational costs.