Modular architectures have been increasingly favored for systems requiring some aspect of flexibility. A growing number of concepts, however, are being developed that have a specific kind of modular architecture in which the entire system is built of self-similar modules. In these systems each module or building block is identical (or very similar) either in external form, or both in form and function. The self-similarity allows for the highest degree of reconfigurability and is emerging to be a consistent choice for systems that need to fulfill multiple roles at different times, evolve easily to respond to new needs, and/or degrade gracefully over time. Since reconfigurability can allow for improving performance, efficiency, reliability, and flexibility of application, its manifestation through self-similar modular architecture for future space systems is studied in-depth in this paper. The systems analyzed in the study range from information processing (such as avionics) to mass transportation/supporting systems (such as spacecraft). Some of the advantages and disadvantages of this architecture are quantitatively analyzed through a case-study of a self-similar modular habitat for Moon and Mars missions. It is shown that self-similar modularity naturally allows for graceful degradation, system extensibility, and learning curve savings in production costs. There are however drawbacks of increased mass and module complexity.