International comparisons of health care expenditures and their determinants have attracted considerable attention since the early 1960s and have since been used widely to compare countries. The impetus for this has been two-fold: firstly, to assess the macroeconomic efficiency of health systems by determining whether different methods of financing and delivering health care have contributed to the control of overall spending levels; and, secondly, to investigate the determinants of the level of health care expenditure. Empirical research has suggested that there is a causal and statistically significant relationship between growth in health spending and growth in gross domestic product (GDP), and that an increase in the latter brings about a proportionately larger increase in the former. This relationship holds even when other potential determinants, such as urbanisation and age structure of the population, are included in the analysis, leading to the conclusion that health care is a luxury good. This paper examines the extent to which this finding is valid from a methodological perspective and how far it assists policy analysis. We argue that there are significant problems in the measurement of health spending and GDP, discuss the methodological problems in the analysis and suggest that the observed relationship between GDP and health spending is unhelpful and almost certainly misleading for health policy development.