A model of cascading failures in complex networks with an emergency recovery mechanism is proposed in this paper, and the cascading dynamics is investigated by running the proposed model on nearest-neighbor coupled network, Erdos-Renyi random graph network, Watts-Strogatz small-world network and Barabasi-Albert scale-free network respectively. New concepts in emergency recovery mechanism and the efficiency of networks are defined. And the effects of the parameters on the network efficiency and failure rate are investigated. Results demonstrate that the increase of the emergency recovery probability would reduce the network efficiency decreasing speed and the failure rate growing speed, and also improve the resilience of the network. And the greater the load capacity of the nodes in the network, the slower the speeds of network efficiency reducing and failure rate growing. Meanwhile, with the decrease of the overload node failure probability, the reducing speed of network efficiency and the growing speed of failure rate would reduce gradually. Furthermore, the changes of the network efficiency and failure rate during the process of cascading failures in different network topologies are analyzed. It is found that the rise of the heterogeneity of degree distribution increases the reducing speed of network efficiency and the growing speed of failure rate. All these results can help analyze the cascading dynamics in complex networks with an emergency recovery mechanism, and may provide a guidance for the controling of cascading failures and protecting against them in real-life complex networks.