Model and component-based development approaches have emerged as an attractive option to deal with the complexity of vehicle software. Using these approaches, we provide a method to estimate and refine end-to-end timing requirements in vehicular distributed embedded systems that are developed by reusing the models of legacy systems. This method leverages on the early end-to-end timing analysis that can be performed at the highest abstraction level during the development of these systems. As a proof of concept, we conduct a vehicular-application case study to show the process of estimating and refining the timing requirements early during the development. The case study is modeled with the Rubus-ICE tool suite that is used for the software development of vehicular embedded systems by several international companies.