This paper summarizes the final results of a study analyzing different Guidance, Navigation and Control (GN&C) architectural approaches for fault tolerance in National Aeronautics and Space Administration's (NASA's) crewed and robotic exploration space systems. GN&C systems were decomposed into simple building block subunits of sensors, computers, and actuators and various forms of subunit interconnection were defined for investigation. The resulting subunit/interconnection construct was used as a top-level abstraction for building candidate GN&C system architectures. This model was implemented using Massachusetts Institute of Technology's (MIT's) Object Process Network (OPN) modeling language in order to more easily enumerate possible architectures and ultimately identify which of these architectures have optimal properties. Dual and triple redundant GN&C system architectures, employing different classes of components, were modeled using the OPN language. The model assumed perfect coverage - 100-percent accuracy in detecting and isolating a failure. Within the constraints of the model, all possible architectures were rigorously enumerated and the weight/reliability trade-offs of crossstrapping components and using more than one type of component were assessed. The study results indicate it is possible to produce nearly all potentially optimal GN&C architectures using generic connections between low-reliability components. The identified optimal architectures reveal a preference to increase GN&C system redundancy of lighter, less reliable components rather than using smaller numbers of more reliable, heavy components.