Aquatic ecosystems are under threat from multiple stressors, which vary in distribution and intensity across temporal and spatial scales. Monitoring and assessment of these ecosystems have historically focussed on collection of physical and chemical information and increasingly include associated observations on biological condition. However, ecosystem assessment is often lacking because the scale and quality of biological observations frequently fail to match those available from physical and chemical measurements. The advent of high-performance computing, coupled with new earth observation platforms, has accelerated the adoption of molecular and remote sensing tools in ecosystem assessment. To assess how emerging science and tools can be applied to study multiple stressors on a large (ecosystem) scale and to facilitate greater integration of approaches among different scientific disciplines, a workshop was held on 10–12 September 2014 at the Sydney Institute of Marine Sciences, Australia. Here we introduce a conceptual framework for assessing multiple stressors across ecosystems using emerging sources of big data and critique a range of available big-data types that could support models for multiple stressors. We define big data as any set or series of data, which is either so large or complex, it becomes difficult to analyse using traditional data analysis methods.