Modern organisations increasingly rely on management science techniques and decision support systems (DSS) to enhance problem solving and decision making. Simulation modelling is one of the most widely applied management science techniques. However, constructing a valid model of a complex real—life system can be a tedious, lengthy and error—prone task, even when powerful model specification techniques, based on computer graphics or natural language processing (i. e. on graphical or textual descriptions of models), are incorporated. This is particulary true when attempting to model systems consisting of vast numbers of interacting elements. The paper proposes an alternative methodology for developing simulation—based decision support tools, wich is based on a DSS combining computer graphics with automatic model generation techniques. A case—study from the railway transportation domain is presented to demonstrate the main features of the system and its impact on decision making and organisational performance.