Agile methods are attracting more and more attention for the development of mechatronic systems. They promise development organisations in companies flexibility and responsiveness, enabling them to react adequately under dynamic and volatile conditions. Originally, agile methods were considered to support software development. However, they appear to be quite suitable for supporting the development of mechatronic systems. The main difference between software development and the development of mechatronic systems was initially seen as being that mechatronic systems have to materialise. Functionalities and properties of mechatronic systems can only be realistically used after they have been manufactured. Physicality is therefore considered to be the most restrictive aspect in the use of agile methods for mechatronics development. More detailed analyses show that the high level of functional integration and the complexity of the development in particular also present a challenge. Division of labour for large complex systems is necessary to be efficient in development. Development tasks are not independent of each other. This requires a rethinking of agile working mechanisms to provide the best possible support for the development of mechatronic systems. Another interesting phenomenon in the use of agile methods is that they were initially taken up from practice. Many companies have experimented with agile methods, which is reflected not least in the fact that large application-focused conferences on agile work quickly became established. However, this is also accompanied by a so-called guru problem. Individual persons or companies demonstrate the successful application of the methods on the basis of best practices. If these are used without reflection, they do not necessarily lead to success, especially if the application contexts differ greatly. Although it is clear from the best practices what works in the application of the method, this is usually only partially explainable because descriptive cause-and-effect mechanisms are lacking. As a consequence, the introduction of the method fails. This means that effective and promising approaches are easily burned. Consequently, the paper was based on the following research questions: How can cause-and-effect mechanisms and means-end relationships of agile development be identified, and how must agile methods be adapted to make them effective for the specific characteristics of mechatronics development and for specific application contexts? The article summarises the findings from a series of studies, from which implications for the use of elements of agile working, in this case prototyping in particular, are derived and explained.