In this paper I test two models of strategic debt behavior. The general idea is that if a government anticipates the possibility of defeat in the next election it will try to use the debt strategically in order to influence the policy of its successor. Previous empirical studies have either rejected the strategic explanation of debt or have not been able to isolate this effect. I argue that these findings are perhaps less surprising since there are several potential difficulties using U.S. or OECD data to test the strategic explanation. One problem with the U.S. data is the scarcity of degrees of freedom while a problem with the OECD data is the pooling assumption. To come to grips with these two problems I use a data set from Swedish local governments. The main advantages of this panel data set are the homogeneity of the sample and the large number of observations from elections, nearly 2000. After controlling for other possible economic and demographic determinants of debt behavior, the main findings of this paper strongly suggest that a right-wing government accumulates more debt during its term of office if it thinks that it will be defeated as compared to when it expects to remain in office. On the other hand, a left-wing government decreases the level of debt the higher the possibility of its defeat. Moreover, the larger the inherited debt the more a newly elected government has to reduce spending and raise taxes. These results are consistent with the predictions from a model developed by Persson and Svensson (1989).