In the Network Flow Interdiction problem an adversary attacks a network in order to minimize the maximum s-t-flow. Very little is known about the approximatibility of this problem despite decades of interest in it. We present the first approximation hardness, showing that Network Flow Interdiction and several of its variants cannot be much easier to approximate than Densest k-Subgraph. In particular, any $n^{o(1)}$-approximation algorithm for Network Flow Interdiction would imply an $n^{o(1)}$-approximation algorithm for Densest k-Subgraph. We complement this hardness results with the first approximation algorithm for Network Flow Interdiction, which has approximation ratio 2(n-1). We also show that Network Flow Interdiction is essentially the same as the Budgeted Minimum s-t-Cut problem, and transferring our results gives the first approximation hardness and algorithm for that problem, as well.