Although the ultimate goal of most current fusion research is to build an economically attractive power plant, the present status of physics and technology does not provide the performance necessary to achieve this goal. Therefore, in order to model how such plants may operate and what their output might be, extrapolations must be made from existing experimental data and technology. However, the expected performance of a plant built to the operating point specifications can only ever be a ‘best guess’. Extrapolations far beyond the current operating regimes are necessarily uncertain, and some important interactions, for example the coupling of conducted power from the scape-off layer to the divertor surface, lack reliable predictive models. This means both that the demands on plant systems at the target operating point can vary significantly from the nominal value, and that the overall plant performance may potentially fall short of design targets. In this contribution we discuss tools and techniques that have been developed to assess the robustness of the operating points for the EU-DEMO tokamak-based demonstration power plant, and the consequences for its design. The aim is to make explicit the design choices and areas where improved modelling and DEMO-relevant experiments will have the greatest impact on confidence in a successful DEMO design.