Network coding has been shown to offer an intrinsic security advantage in terms of node-based wiretapping attack. Previous literature explored the security advantage by defining and investigating the security level at an arbitrary intermediate node. As pointed out in other studies and observed by us, in order to achieve the security level at all intermediate nodes, it is necessary to have an admissible link encryption scheme to prevent all links from eavesdropping. “Admissibility” here promises that an adversary cannot obtain more meaningful (decoded or decodable) information about the original message than that disclosed at the node he controls. In this article, we study a similar problem but focus on link encryption. We propose an admissible link encryption scheme LNCLE; in conjunction with LNCLE, linear network coding achieves its security level at all intermediate nodes. In comparison with a trivial link encryption scheme (TLE) which is based on classical link encryption, LNCLE is more efficient in the sense that both key size and computation cost are greatly reduced.