There have always been differences of view on what poverty means in conceptual terms, and even greater differences on how to measure it. These differences span a broad spectrum of normative and ideological positions and raise a number of technical issues surrounding the statistical measurement of poverty. This paper explains the role of poverty research and the value of a poverty line, while acknowledging that limitations exist with the current instruments. It argues that any poverty measure must include two key ingredients of poverty – the idea that resources are inadequate to meet basic needs and the notion that needs can only be defined relative to prevailing community attitudes and standards. Survey results are used to support the view that most Australians see poverty in subsistence terms, but this does not contradict the idea of relativity, since subsistence is itself a relative concept. The principal arguments are illustrated using data from the 1998-99 Household Expenditure Survey to estimate poverty on the basis of incomes, expenditures and a combination of a conventional income measure with additional data on hardship. The poverty rate is shown to be sensitive to which measure is chosen, both in aggregate and for specific groups in the population. However, all measures show that the group with the highest incidence of poverty is sole parent families, and that there is a strong association between joblessness and poverty, with full-time employment being required to escape poverty. The poverty rate among the aged is high when the conventional (income-based) measure is used, but far lower aged poverty rates are produced by many of the alternative poverty measures. State differences in income poverty are substantial but become much smaller when a deprivation-adjusted poverty measure is used.