Rail vehicle models have become increasingly complex, posing challenges in extracting insights using traditional model representations as they require numerous iterations to achieve a satisfactory solution. This complexity leads to high computational and time costs and possibly resulting in inefficient vehicle design. To alleviate these limitations, network models are proposed as an alternative representation in this paper. These models enable the analysis of structure, behaviour, and patterns of interactions, facilitating an understanding of knock-on effects across disciplines and subsystems. The terminology, benefits, and capabilities of network theory in early-stage vehicle design are presented in this paper, along with the aspects to consider and methods for developing network models. The applicability of network theory metrics and algorithms is demonstrated using a railway traction system example. Results indicate that the proposed representations can capture complex system knock-on effects across disciplines and subsystems.