The interconnection network used in a multiprocessor system provides an effective mechanism for the data exchange between the processors and is one of the dominating factors of performance of the system. The fault tolerance of an interconnection network can be measured by the probability that the network keeps functional in the presence of failures. The authors examine the fault tolerance of four regular interconnection networks (i.e. hypercube, crossed cube, Moius cube, and locally twisted cube) with the probabilistic approach. Experimental results show that all these interconnection networks display excellent fault tolerance. Moreover, the crossed cube enjoys the highest probabilistic of connectivity among them.