We employ a rich sample of individual tax returns data to simulate alternatives to the income tax reform introduced in Germany by the governmental coalition of Social Democrats and Greens in 1998. We characterize three reforms that would have been fiscally equivalent to the actual one: a distributionally neutral tax reform, one with a maximal basic allowance, and a flat tax. By comparing the individual tax burdens under each alternative, we simulate majority voting on the tax reform. The actual reform loses against both the distributionally neutral reform and the one with maximal basic allowance; however, it wins against the flat tax. The Condorcet winner turns out to be the reform with maximal basic allowance.