The evidence found in most studies suggests a strong positive relationship between health care expenditure and gross domestic product. However, this evidence weakens with respect to the actual value of the income elasticity. There are two possible sources of these discrepancies, the use of arbitrary deflators and specification errors. We find that health PPP cannot be taken as a 'universal' price index. The problem is that its components do not move together. Nevertheless, we derive a 'universal' health price index from a dynamic system in which its components share both short and long run co-movements. The omission of relevant explanatory variables seems to be the main cause of the discrepancies. We confirm that there exists a strong positive relationship between per capita health care expenditure and per capita GDP. However we estimate a long run income elasticity at or around unity, although it is greater than unity for the countries with lower per capita income (Spain and Ireland). The results for income elasticity are the same regardless of whether health care expenditure is converted using the GDP PPP or the 'universal' health price index. The importance of non-income variables is also confirmed, in particular the relative price of health care. We find that relative price has a strong rationing effect on the quantity of health demanded and has no effect on the expenditures.