This article considers changes in poverty rates under the Howard government. We also make three methodological contributions. We consider the statistical significance of the estimated changes in poverty. We propose a decomposition technique that reconciles the trends in absolute and relative poverty. We also use 'poverty profiles', which illustrate sensitivity to alternative poverty lines. We find decreases in absolute poverty and increases in relative poverty, both of which are statistically significant over a range of poverty lines. At a poverty line equal to half of the median income, the increase in relative poverty is statistically significant for all people and borders on significance for children.