The ecosystem modelling complex ‘Ecopath with Ecosim’ has been implemented extensively in the field of marine science; however, despite its widespread application, descriptions of its functionality remain arcane in the literature. This study conducts an evaluation of the software’s prediction capacity using eight published Ecopath models. Response of six ecosystem-status indicators to four basic input variables to which imprecision had been added was investigated. Kempton’s Q Index and total system throughput emerged as the most consistently responsive parameters. Moreover, input biomass was identified as a ‘high-leverage’ parameter, its influence on outputs being greater than that exerted by any other input variable. This study constitutes one of the first comprehensive investigations of the response of selected outputs to imprecise input values, and provides sufficient basis to warrant a sensitivity assessment of the software, as well as introduction of a dedicated tool to perform such a task within Ecopath with Ecosim.