Industry is constantly looking for new developments in software for use in increasingly complex computer applications. Today, the development of component-based systems is an attractive area for both Industry and Academia. The systems we focus on in this thesis are embedded computers, in particular those in automotive systems. A modern car incorporates several embedded computers that control different functions of the car, e.g., anti-spin and anti-lock breaks. The main purpose of this thesis is to investigate how component technologies for use in embedded systems can reduce resource usage without compromising non-functional requirements, such as timeliness. The component-technologies available have not yet been used extensively in the vehicular domain. To understand why this is the case we have conducted a survey and performed evaluations of the requirements of the vehicular industry with respect to software and software development. The purpose of the evaluation was to provide a foundation for defining models, methods and tools for component-based software engineering. The main contribution of this work is the implementation and evaluation of a framework for resource-efficient mappings between component-models and real-time systems. Few component technologies today consider the mapping between components and run-time tasks. We show how effective mappings can reduce memory usage and CPU-overhead. The implemented framework utilizes genetic algorithms to find feasible, resource efficient mappings. We show how component-models designed for resource constrained safety-critical embedded real-time systems can use powerful compile-time techniques to realize the component-based approach and ensure predictable behaviour. Further, we propose a resource reclaiming strategy for component-based real-time systems, to decrease the impact of pessimistic execution time predictions. In our approach, components run in different quality levels as unused processor time is accumulated.