Militarized state-building interventions (MSIs) must fulfill two often conflicting goals. At the time of withdrawal the intervenor must leave in place a state able to survive on its own and govern its territory. States only intervene in other states, however, when they aspire to change the policy of the target in ways they prefer. In attempting to balance these objectives, the intervenor ‘pulls’ policy in its preferred direction by supporting a less popular leader at the cost of leaving behind a state that is no more likely to survive over time than its peers. We test our theory and find evidence for this trade-off by examining all MSIs by great powers and IOs in failed states from 1956 to 2006. Consistent with the theory, we find that MSIs do not on average have any significant effect on state survival. We also find that MSIs that move the target state’s policy closer to that of the external power have a negative effect on survival, but interventions that do not result in a change in policy do not. This argument and finding temper the optimism of much of the contemporary literature on international interventions. Potential intervenors face a stark trade-off. If they draw the policy of the failed state towards their own preferences, then that state will be more likely to fail again in the future.