The majority of innovations in modern automotive systems are driven by software. Based on the research results of the software engineering community over the past decades we are able to develop software systems with immense complexity. However, concomitant withthese increases in complexity, the quality demands on modern cars also appear to be ever-growing. Probabilistic properties are commonlyapplied to specify these quality demands and are especially suitable for performance, reliability, safety, and availability requirements. This talk will present some recent advances on verification of probabilistic properties at design time, as well as some approaches to optimise automotive systems and software architectures with respect to these properties. Afterwards, monitoring approaches are introduced that allow run-time verification. Complemented with time-series analysis techniques, this allows dynamic detection and proactive resolution of problems in modern cars.