We present the metrics for evaluation of network alignment, which refers to the process of identifying node (and edge) correspondence across multiple networks. These metrics are defined in the context of the DARPA Modeling Adversarial Activity program, where a key technical area is to develop novel mathematical techniques to merge networks from various sources. Specifically, two metrics are presented for evaluation: vertex-based metric and edge-based metric. The first metric is defined based on the assumption that ground-truth alignment is given between vertices across the channels. The idea is to evaluate the approximation of the output from network alignment algorithms with respect to the ground truth. Networks alignment to address the node (and edge) correspondence problem The second metric is defined based on matching edges, which also provides and alternate view on assessing the alignment confidence in scenarios where ground truth alignment is not available. Examples on evaluating different network alignment outputs based on our metrics will be presented in the paper.