Modern vehicle electrical architectures are extremely complex and based on large number of Electronic Control Units (ECUs) integrated with hardwiring or a network technology such as Controller Area Network (CAN), Local Interconnect Network (LIN) or FlexRay. Even a simple system such as a driver's door can contain a large number of control functions such as electric windows, electric mirrors, seat control switches and therefore a lot of internal wiring is required within the door system. To reduce cost, weight and complexity of the wiring harness, it is usual for some combination of CAN, LIN and hardwired connections to be used for electronics integration. The key problem with this is how to assess the potential for cost and weight saving of candidate architectures at the very early stage of the design process and therefore how to choose a particular electrical architecture. At the very early stage of the design process there is likely to be very little information about the cost and weight of the system available.The objective of the research discussed in this paper was to develop a methodology for comparing candidate electrical architectures based on both cost and weight design targets with the minimal amount of information. At the very beginning of the design of an automotive electrical architecture it is desirable to be able to make decisions about partitioning of an architecture between in-vehicle networking technologies and hardwired integration using coarse information. Two case studies are described. The first is concerned with the design of a harness for a mass produced passenger car. The second is concerned with the design of a body control harness for a niche sports car manufacturer. For both case studies it is shown how historical data concerning the cost and weight of the original wiring harness is analysed and applied to a number of candidate architectures utilising electrical integration based on the Local Interconnect Network (LIN) technology. As well as the case studies, the generic process is described and discussed how it could be applicable to the assessment of other types of automotive electrical architectures.