The traditional starting place for a study of tax reform is a definition of an ideal tax base, which is then adjusted in light of additional issues. After arguing briefly that this is not a good starting place, the essay reviews the optimal taxation literature inferences for tax base policy, focusing on three questions: --- If there is annual non-linear (progressive) taxation of earnings, how should annual capital income be taxed - not at all, linearly (flat rate, as in the Nordic dual income tax), by relating the marginal tax rates on capital and labour incomes (as in the US), or by taxing all income the same? --- If there is annual non-linear taxation of earnings, should there be a deduction for net savings? --- If there is annual non-linear taxation of earnings, is it worth having a more complex tax structure, particularly age-dependent tax rates? Would greater use of age-dependent rules in capital income taxation be worthwhile? The essay presents the Atkinson-Stiglitz and Chamley-Judd results that capital income should not be taxed, but concludes that the required conditions are too restrictive and not robust enough for policy purposes. Hence there should be some role for including capital income as a part of the tax base. The essay discusses some empirical underpinnings for two key elements in the conclusion - differences in savings propensities and the shape of earnings (and uncertainty about earnings) over the lifetime. The conclusion that capital income should be taxed does not lead to the conclusion that the tax base should be total income, the sum of labour income and capital income. The chapter leans toward relating marginal tax rates on capital and labour incomes as opposed to the Nordic dual tax. Also examined are age-dependent taxes (for example different taxation of earnings for workers of different ages). Keywords: Earnings tax, Capital income tax, Income tax, Age-dependent tax, Horizontal equity, Meade Report. JEL Classifications: H20, H21, H