This analysis of formula apportionment compared to the current system is based on the observation that income shifting has two sources, intangible income and debt. The analysis also recognizes that a major goal of the transfer pricing or income allocation system is to preserve the tax neutrality between arm’s length and related party transactions and between multinational and single jurisdiction companies. It therefore develops a model that highlights these features. Both separate accounts (SA) and formula apportionment (FA) distort behavior but along different margins. Under SA, companies have an incentive to shift high-tech activities and to manipulate transfer prices. Under FA, companies do not manipulate transfer prices but they have an incentive to shift routine activities abroad and to change the degree to which they depend on outside suppliers. Simulations based on the model indicate that FA has no clear advantage over SA even when the model assumes that an unrealistically large amount of resources are devoted to tax planning under SA. Furthermore, straightforward changes could be made in SA that would result in substantial improvements without resorting to full-fledged FA. We also examine the complicating role of financial assets under FA and how ongoing R&D is implicitly allocated. The conceptual basis for the conventional formulas are discussed, particularly ones based on sales. Finally, a static, no behavioral change, estimate of the effect of FA on the tax liabilities of US multinational corporations is presented for 1996 and 2004. The static estimate for 2004 suggests a potentially large revenue gain, but the simulations show that tax revenues under FA and SA are similar when behavioral responses are taken into account.