The increase demand for flexibility, spontaneity, adaptability, independence of a wire infrastructure, and easy-managed architectures are driving a revolutionary change in the network community research. The number of proposals in the literature related to technologies like ad hoc, and more recently, sensor networks witnesses for the interest on the deployment of distributed, autonomous, spontaneous, and self-organizing systems. Distributed hash tables (DHTs) have been largely adopted as a useful substrate to the design and specification of self-organizing systems. The functionality of decoupling identification from location, and of providing a general mapping between them, have made the DHT abstraction an interesting principle to be integrated in network-level routing protocols. Although improving scalability, this integration also introduces new challenges for the network architecture specification. In this paper, we examine the main components related to the design of network architectures having DHT interactions in mind. We survey the recent proposals that attempt to deal with DHT-based self-organizing systems, and discuss their differences according the way in which the DHT functionalities are applied. It is verified that initial decisions of the designers about priorities of different issues and requirements are reflected on proposals’ strengths and weakness. We conclude our investigations by presenting a discussion about the proposals’ more significant features. Finally, based on our concluding observations, we also discuss a new and promising approach for the deployment of a distributed and self-organizing system.