The new clock synchronization mechanism for distributed systems proposed in this paper is based on the interactive convergence algorithm, CNV, but is much more reliable in the presence of faulty nodes in a partially connected distributed system such as the Hypercube. The authors give a formal proof of the correctness of their algorithm under well-defined working conditions. The proposed solution is very attractive for the synchronization problem in distributed systems, since simple hardware (a high frequency quartz oscillator, a programmable counter/divider, and a counter to count the divided clock) and reworked synchronization software give clock skews that are several orders of magnitude smaller than those obtained using the software-only solution. Compared to the cost of a fully hardware-assisted synchronization, the authors’ approach may be viewed as optimal.