Natural disasters often lead to large-scale transportation network destruction. Besides an evaluation of the damages, an urgent issue would be how to restore the transportation network's basic capability in a timely manner after the occurrence of a natural disaster. This research aims to establish a network reliability evaluation process for solving the problem of whether the disaster area network capacity can satisfy the basic demands of evacuation, further for pointing out critical links where capacity may be degenerated by natural disaster. We can use the network reliability evaluation process to allocate resources on those critical links for restoring the emergency evacuation network. This research is discussed in two stages. First, Maximal Flow Problem is applied to evaluate the reliability of disastrous network for computing the maximal supply of the network. Second, an intention is given to restore the network for achieving the most efficient evacuation based on the reliability evaluation. This research belongs to the network design field and the subject can be formulated as a bi-level programming model. The upper level of the model is to minimize the system’s total travel cost under limited network restoration resources. The lower level is to analyze the destroyed network and allocate resources to those critical links for restoring the emergency evacuation network under link capacity constraints by considering victim evacuation route choice and O/D choice behaviors. After solving the problem, the variational inequality sensitivity analysis method, generalized inverse approach, and gradient projection method are adopted to provide the solution algorithms for the bi-level programming model. The numerical test results indicate that the Stackelberg equilibrium solution of the formulated bi-level programming model does exist. More significantly, the network reliability evaluation, optimal network restoration planning, disaster victim emergency evacuation route planning can be achieved.