The speed and reliability performance of computers are significantly influenced by the underlying interconnection structures. Combinatorial methods have been used to compute the reliability of interconnection networks. However, time-dependent reliability analysis requires the use of a stochastic process such that the reliability performance can be formulated more accurately. This paper discusses a new reliability modeling approach, i.e., the Stochastic Dataflow Graph (SDFG) Models for the reliability modeling of interconnection and computer networks. Examples of interconnection networks and computer networks are analyzed based on the SDFG models (including continuous time SDFG, and SDFG with repair parameters) to illustrate this new reliaility modeling and analysis approach. The proposed reliability analysis models provide system engineers with a useful tool for evaluating the reliability performance of interconnection and computer networks.