What is the appropriate basis for the comparison of the costs of families with and without children, and how is this relevant to social policy? This paper discusses these issues and derives some simple equivalence scales for the relative costs of children. The method used, originally due to Rothbarth, rests on the comparison of expenditures on adult goods in families with and without children. Data from the 1984 Household Expenditure Survey suggest average costs of children (relative to couples) of around 20 percent. Whilst the data are suggestive of strong economies of scale and different impacts of child costs on mothers' and fathers' consumption, conclusions on these issues must await a larger sample.