In the context of impacts research, detection and attribution exercises evaluate whether aspects of human and natural systems are changing in response to the impetus of climate change. Concepts and methods for detection and attribution have been established in the physical science community, evaluating changes in the climate system due to anthropogenic forcing. In contrast, a conceptual framework for detection and attribution of climate change impacts that is consistent and applicable across disciplines is still lacking. In an attempt to overcome this methodological deficit, this paper outlines the major challenges involved in and provides workable definitions of detection and attribution in the context of impacts. Reaching beyond the current focus of the literature on hydrology and ecological effects, it sets a focus on challenges that are inherent in the particular dynamics of human and managed systems, including non-stationary baseline behaviour, multiple drivers, and active adaptation.