<p>Model-Based Systems Engineering (MBSE) is an approach to managing the complexity of modern cyber-physical systems, including automotive systems. In the domain of automotive engineering, it is common for engineers to use a variety of languages at various levels of abstraction, to provide diverse and concrete perspectives of a system. However, a significant incompatibility challenge arises due to weak or nonexistent integration among these languages. This challenge leads to redundant effort and compromised traceability and can hinder the automation of the systems and software development processes. In a previous study, we proposed a model-to-model (M2M) transformation that maps SysML system models into AUTOSAR software models. The transformation considers the client–server and sender–receiver communication patterns supported by AUTOSAR. In this article, we elaborate on the transformation approach in more detail to present a comprehensive account of the transformation, as well as to provide guidelines to practitioners on system design constraints needed to enable the transformation. We also discuss the limitations of the implementation from the perspective of the characteristics of model transformation languages and show how the limitations can be overcome by using a transformation language with different capabilities. We evaluate our transformation approach by applying both implementations to three real-world case studies with different communication patterns and present a comprehensive comparison between the two implementations. Based on our findings, we conclude that the presented model transformation approach is effective in bridging the gap between models at the systems architectural and software architectural levels, and that the previous limitations can be addressed by our new implementation.</p>