Railway suicide constitutes a limited but significant fraction of all suicides, and a majority of all railway-related deaths, in Sweden as well as in many other countries. The aim of this thesis is to develop the scientific basis for the prevention of railway suicide, with special regard to knowledge and strategies within the reach of the railway system owners themselves. An initial descriptive epidemiologic study showed that railway suicide episodes cluster in densely populated areas, and that victims often seem to await approaching trains for some time on or close to tracks before being hit. This means that preventive efforts can be prioritized to geographically limited areas and that there exists a time margin for early detection, forewarning, braking, rescuing, and the like. Secondly, a theoretical study was conducted in order to derive an overview of theoretically possible barriers aimed at preventing railway suicide. The approach is based on the comparison and synthesis of theories and models from suicidology and safety sciences applied to the railway suicide phenomenon. This led to a list of possible barriers, also intended as a checklist. This was followed by a multiple case study, where a number of authentic cases were analysed in-depth on the basis of investigation reports by the police and railway officials, plus complementary researcher observations from the actual sites of the suicide occurrences. The study revealed some deficiency in current investigation and reporting routines from a preventative viewpoint and indicated the potential for a broader and more systematic collection of prevention-relevant information in immediate connection with the events. Finally, a focus group study, involving safety practitioners from both the railway and other sectors, confirmed a general understanding and acceptance among practitioners of the theoretically derived principles, and a readiness to broaden the scope of railway safety to include approaches to prevent suicide. In conclusion, it should be possible to influence the occurrence of railway suicides by means available within the railway transport system itself. A spectrum of possible alternative barriers is identified, as well as an, as yet, underutilized potential in the form of more systematic data collection and learning routines from actual events. Last but not least, there is a readiness among practitioners to incorporate these principles in regular railway safety work.