The identification and prioritization of multiple watershed stressors and the corresponding development of optimal management strategies remains a challenge in ecological risk assessment. Eco-epidemiological analysis of archival environmental spatial databases integrates available biological, physical and chemical information to generate hypotheses based on stressor-response data relationships. A geographic information systems-based technique incorporating Bayesian weights-of-evidence analysis and weighted logistic regression (WOE/WLR) developed for and currently utilized in minerals exploration was extrapolated to the eco-epidemiological analysis of aquatic ecosystem data. Case studies within state of Ohio (USA) and England and Wales were conducted to demonstrate a method proof-of-concept in the context of various concepts relevant ecological risk assessment such as biological endpoint selection, geographic scale, land use characterization, and temporal variability. Analysis results were communicated as quantitative estimates of stressor influence and impairment probability maps with significant model fit to observed data. Stressor identification, influence and estimated impairment probability varied across levels of biological organization, study area extent, land use intensity, and seasonal conditions. WOE/WLR results yielded quantitative evidence for the importance of consideration of environmental complexity in ecological risk assessment, and were consistent with both biological plausibility and examples from the literature. Validation of WOE/WLR stressor hypotheses was additionally evaluated by cross-comparison to an independent eco-epidemiological database method and to various field-based assessments. A methodological framework for the development and application of WOE/WLR for eco-epidemiological research is proposed to enhance screening-level (Tier 1) ecological risk assessment.