Breach remedies can be used to protect specific investments and are therefore a remedy against holdup. Yet some commonly used remedies are predicted to provide too much protection, thereby inducing overinvestment. Two motives drive this prediction: the insurance motive and the separation prevention motive. This paper presents results from an experiment designed to test whether these two motives show up in practice. In contrast to previous experiments the focus is on a setting where ex post renegotiations are possible. Our results indicate that also in this case the insurance motive and the separation prevention motive are at work, as predicted. A second main finding is that there seems less need for sophisticated breach remedies based on compensatory money damages than is suggested by theory.