Society needs believable information about poverty for its self-understanding and for rational political exchange and social policy. It is an ambition of poverty research to provide society with that information. In spite of decades of effort and volumes of excellent research, sociologists and economists have had only limited - if any - success in achieving this aim. There is a view that for this purpose poverty research needs to seek a methodology for the measurement of poverty. Here, it is suggested that the object of measurement should be 'the problem of poverty'. Basic conventions in definition and measurement are re-examined, and eight recommendations for measurement are developed. The main results are the rejection of the very notion of a poverty line that divides the population into 'the poor' and 'the not-poor' as the commanding instrument for measuring the problem of poverty and the recommendation against the use of approaches in which poverty is established from relative information only. The alternative, it is suggested, is a social indicator approach, grounded in the principle of poverty as freedom denied, and encapsulating degrees of deprivation in the form of destitution, want, and risk.