Recent technology advancements and a global shift in the transportation paradigm have led to an accelerated development of, and demand for, highly autonomous vehicles (AV). However, there are still a number of implementation obstacles due to unanswered safety, ethical, social, political, regulatory, and legal issues. Currently, there is still lack of a well-defined assessment method that can verify the safety performance of AV. This paper attempts to develop an effective method for safety and reliability assessment of fully automated (level 5) vehicles. The restricted analysis of level 5 AV allows a simplification of the model, since the human operator is by definition eliminated from the control structure. System Theoretic Process Analysis (STPA) is adopted to provide a qualitative and comprehensive evaluation, resulting in a system design precisely tailored to the safety and reliability requirements of AV. To conduct STPA, a control structure including the complex interplays of its various components was designed, based on state-of-the-art functional architectures used by the automotive industry. Together, the different control loops running simultaneously, and their comprehensive analysis enable real-time fault detection and an understanding of how the errors spread and propagate across the system. The proposed model permits car manufacturers to improve their system design, and assists governments in shaping the appropriate policies and regulations for fully AV.