The paper looks at the relative importance of state benefits, employer benefits and private sources in supporting early retired men. Using data from various administrative sources, the General Household Survey and the Family Expenditure Survey, it shows how the increase in early retirement which occurred between 1979 and 1986 went hand in hand both with a greater reliance on means-tested benefits and a greater reliance on employer benefits. Although the income of the early retired was well above the minimum accorded by the ‘income support’ system, there were major differences between subgroups of the early retired—those dependent solely on state benefits and those with other sources of income, private and personal—and these differences have become more pronounced over time. Finally, the paper looks at the total costs of early retirement: to the state, to employers and to the early retired themselves. Total costs ballooned between 1979 and 1986. Much of the extra expenditure on benefits fell on employers, but the early retired themselves, through accepting a considerable reduction in their income, bore the bulk of the costs of early retirement.