Cyber-physical systems (CPS) are frequently characterized by three essential properties: CPS perform complex computations, CPS conduct control tasks involving continuous data- and signal-processing, and CPS are (parts of) distributed, and even mobile, communication systems. In addition, modern software systems like CPS have to cope with ever-growing extents of variability, namely variability in space by means of predefined configuration options (e.g., software product lines), variability at runtime by means of preplanned reconfigurations (e.g., runtime-adaptive systems), and variability over time by means of initially unforeseen updates to new versions (e.g., software evolution). Finally, depending on the particular application domain, CPS often constitute safety- and mission-critical parts of socio-technical systems. Thus, novel quality-assurance methodologies are required to systematically cope with the interplay between the different CPS characteristics on the one hand, and the different dimensions of variability on the other hand. This thesis gives an overview on recent research and open challenges in model-based specification and quality-assurance of CPS in the presence of variability. The main focus of this thesis is laid on computation and communication aspects of CPS, utilizing evolving dynamic software product lines as engineering methodology and model-based testing as quality-assurance technique. The research is illustrated and evaluated by means of case studies from different application domains.