Carbon taxes introduce potentially uneven cost burdens across the population. The distribution of these costs is especially important in affecting political outcomes. This paper links dynamic overlapping-generations and microsimulation models of the United States to estimate the initial incidence of a carbon tax across states. Geographic differences in incidence are driven primarily by differences in sources of income. Differing patterns of energy use also matter but are relatively less important. The use of the carbon tax revenue plays an important role, particularly in determining how different income sources are affected, as: (1) using carbon tax revenue to cut capital taxes disproportionately benefits states with large shares of capital income; (2) returning the revenue via lump-sum transfers favors relatively low-income states; and (3) returning the revenue via cuts in labor taxes provides a relatively even distribution of cost across states. In general, geographic differences in incidence are substantially smaller than the differences across income groups.