The first part of this paper develops a theoretical model of fiscal equalisation to derive an optimal equalisation formula that has general applicability for federations. If vertical equity is achieved by the central government and horizontal equity by interstate migration, the welfare-maximising role of fiscal equalisation is to support an efficient distribution of different labour types across states. The theoretical model further develops work by Boadway and Flatters and by Albouy with some Australian-oriented extensions. Under optimal equalisation, full equalisation is applied when a state government achieves a fiscal advantage by using source-based taxes to raise revenue from non-residents (e.g. mining royalties) or from spreading its fixed costs over a large population. However, efficiency also dictates that equalisation is limited to fiscal advantages arising from differences in state population composition when applied to revenue raised from residents (e.g. via labour or consumption taxes) or to the variable costs of government. Simplifying assumptions are discussed. The second part of this paper applies the optimal equalisation approach to Australia using, as a base, the Commonwealth Grants Commission (CGC) assessments for 2017/18 and 2018/19. 2016 Census data is used to measure the impact of each state’s population composition on its ability to raise revenue from residence-based taxes for use in applying limited equalisation. The gain in consumer welfare of moving from the current Australian full equalisation system to optimal equalisation is estimated. Policy alternatives that are found to be inferior to the current full equalisation system include no equalisation, half-equalisation and the donor relief equalisation schemes recently recommended by the Australian Productivity Commission.