We experimentally study pre-commitment power's role in credible threatening in adversarial bargaining. A proposer demands payment from the responder, and if the responder rejects to pay, a conflict starts. Before the proposer makes the demand, she chooses the intensity of the conflict. A higher intensity makes both players worse off. The proposer is only partially committed to the intensity; if the conflict starts, she can reduce the intensity up to a fraction of it. In equilibrium, the proposer gains bargaining power by choosing an overly large intensity, which puts both players in a position of having a loss if the conflict starts. The data are qualitatively consistent with the theory. If pre-commitment power allows it, subjects choose a high intensity and tie their hands for the conflict. However, they chose a lower intensity than predicted. Although they gain bargaining power, it is not reflected in the payment they demand from the responder.