Troubleshooting trees are traditionally used to guide technicians through the process of identifying the cause of vehicle problems and solving them. These static trees can successfully visualize complex information. However, for modular vehicles, the trees become difficult to create and maintain due to the numerous different configurations of vehicles that can be constructed. These issues can be overcome by using a model-based approach. This paper describes a prototype tool for guided troubleshooting and shows its application to a selective catalytic reduction system used in many heavy vehicles. The troubleshooting tool guides the technician through the troubleshooting process by presenting the most likely fault candidates and recommending the most useful actions to perform. The list of candidates and recommendations are updated continuously to reflect the outcomes of past actions. An important aspect of the tool is that it removes the user’s need to search and browse for diagnostic information. Instead, relevant information is pushed to the user given the current situation. The troubleshooting system is cloud-based which enables it to always have access to updated data such as product information data and vehicle data pushed out over a remote connection. The performance of the troubleshooting tool as well as the modeling process has been evaluated and compared to existing traditional approaches. The evaluation shows strengths and weaknesses with the troubleshooting system and generally shows an improved performance.