The tax impact on capital structure choice estimated in existing empirical studies depends substantially on the applied tax measure. Among the existing approaches, tax rate measures relying on a simulation of future income have the particular advantage of taking detailed account of the possible development of future income. This advantage has lately been demonstrated in a comprehensive meta-study by Feld/Heckemeyer/Overesch (2011). Existing simulation studies, however, apply three different techniques to modelling future income coming to differing results for the marginal tax rates and the tax optimal debt/equity-ratio: Graham (1996) relies on a firm-specific random walk, Blouin/Core/Guay (2010) apply a cluster-specific mean-reversion process, whereas Graham/Kim (2009) use a firm-specific autoregression-model. Against this background, the present paper aims at analyzing the characteristics of the different tax rate measures proposed in these studies and suggesting an own simulation approach based on these findings.