Physical assaults are a pertinent problem of society (e.g. Kajs, Schumacher, & Vital, 2014; Tiesman, Hendricks, Konda, & Hartley, 2014). One strategy in order to prevent violence is to strengthen the capacities to defend oneself (Koss, 1990), which is the scope of various self-defence programs and systems. Self-Defence skills are being taught throughout the world targeting different users. Even though potential effects on daily behaviour include the use of self-protective strategies, it is important to document if individuals learn the skills taught in self-defence classes and if they are able to perform the skills when these are warranted (Gidycz & Dardis, 2014). Very limited behavioural or self-reported information exists about the mastery and use of self-defence skills. This leads to uncertainty in which approach to use regarding the actual application of skills in a real world settings. This lack of identifying key moderators for program effectiveness provides an opportunity for ideologically loaded arguments about which self-defence system or approach can be applied best. Consequently, regarding the effectiveness of self-defence approaches, even in professional domains like the law enforcement sector, there is no consensus on which approach is most effective. The methodological problem in investigating effectiveness of skill application in an experimental setting is the gap between safety and reality. In real world incidents high physical, cognitive and emotional demands are posed on the individual. Due to ethical concerns (e.g. safety of participants) the level of demand can only partially be applied in experimental settings, making it hard to identify key moderators of program effectiveness. The presentation explores how and to what extent simulation designs can bridge the gap between reality and safety, providing a basis for discussion regarding the investigation of self-defence training effectiveness.