Reflective approach for systematic analysis of risks due to human factors in decision support systems for ill-structured situations is proposed. This approach allows considering not only and not so much the content of the process of new knowledge generation by means of computer-aided technology, but also various roles of people involved in this process, with their cognitive means and cognitive transformations of knowledge. During several years the proposed approach has been verified and refined while solving real-world and experimental problems, mainly with modeling problem situations by formal means of cognitive maps. That enabled disclosing a number of risk factors practically significant in decision support systems for ill-structured situations, those stemming both from decision-making personnel and computer-aided technology designers.