This paper presents a structured method, based on practical experience, to derive critical success factors using a business game called ‘scenario playing’. It aims to show why this method is superior to other well-known focusing techniques or group evaluation methods. The scenario playing business game consists of submitting to managers a sequence of reports, one after the other. Each report (called a scenario) contains randomly generated outcome values for some parameters of the business (called candidate critical success factors) which the manager deemed to be most important for the business at hand. After reviewing each report i.e. scenario, the manager has to give a value judgement about it, like ‘good’, ‘bad’ or ‘indifferent’, reflecting his perception of how the business would be doing if the report was real. A statistical method is used to evaluate these answers and deduct from them which parameters are the most important for the manager who played the given session of the game. The paper contains a short outline of the statistical evaluation (learning) algorithm used by the author and the description of a software tool that contains an implementation of this technique. Finally, the experience gained by using this tool and approach to derive critical success factors on different projects is discussed.