Abstract Background: Complexity received increasing attention from researchers in integration and evaluation studies. Complex adaptive systems (CAS) are the most prominent formulation of complexity used in programme evaluations. However, there remain significant theoretical and conceptual barriers to using complexity as an explanatory model in social sciences, and thus in applying it successfully in integration and evaluation studies. Method: Discussion paper outlining the potential uses and risks of complexity theory for studying integration programmes in health and social care and programme evaluations in general. The paper proceeds by synthesising the work of various critics and proponents of complexity theory in the social sciences and evaluation theory. Results: Complexity offers opportunities and risks to social scientists working in programme evaluations and integration studies. The opportunities are defined by additional modelling and verification/falsification of possible cause and effect links in programme settings. The risks, on the other hand, are twofold. Social scientists may use complexity as a shorthand for as yet insufficient understanding of the contexts under examination, or they mistake it for an explanatory device without testing its potential to explain. The second risk emerges as a result of the nature of complexity and its role in natural sciences. Assigning complexity an explanatory role may prevent further investigation of a given setting that may reveal that complexity is insufficient to understand what is going on. Implications for practice: Researchers should make clear how they have operationalised and measured the various features of the complexity model to allow robust verification of the evidence. Scholars should also assume that complexity as defined by the natural sciences is philosophically and epistemologically problematic when transferred into the realm of social sciences that largely operate with concepts informed by the paradigm of understanding social behaviour.