The ecosystem services (ES) approach offers an integrated perspective of social-ecological systems, suitable for holistic assessments of mining impacts. Yet for ES models to be policy-relevant, methodological consensus in mining contexts is needed. We review articles assessing ES in mining areas focusing on freshwater components and policy support potential. Twenty-six articles were analysed concerning (i) methodological complexity (data types, number of parameters, processes and ecosystem-human integration level) and (ii) potential applicability for policy development (communication of uncertainties, scenario simulation, stakeholder participation and management recommendations). Articles illustrate mining impacts on ES through valuation exercises mostly. However, the lack of ground- and surface-water measurements, as well as insufficient representation of the connectivity among soil, water and humans, leave room for improvements. Inclusion of mining-specific environmental stressors models, increasing resolution of topographies, determination of baseline ES patterns and inclusion of multi-stakeholder perspectives are advantageous for policy support. We argue that achieving more holistic assessments exhorts practitioners to aim for high social-ecological connectivity using mechanistic models where possible and using inductive methods only where necessary. Due to data constraints, cause-effect networks might be the most feasible and best solution. Thus, a policy-oriented framework is proposed, in which data science is directed to environmental modelling for analysis of mining impacts on water ES. © 2018 by the authors.