The purpose of this study is to defend the view that education should be evaluated in terms of the capability to achieve valued functionings, rather than mental satisfaction or resources. In keeping with Amartya Sen's capabilities approach we argue that mental satisfaction provides an inaccurate metric of well-being because of the phenomenon of adaptive preferences. Equally, resources cannot be used as a metric of well-being because of inequalities in the ability to convert income and commodities into valued functionings. Hence, interpreting education as a means to create human capital is also impoverished because it evaluates education solely in terms of the accumulation of resources. In order to provide evidence in support of the human capabilities approach we statistically examine the channels through which educational attainment affects the health functionings implied by life expectancy. Using panel data analysis for 35 developing countries for the years 1990, 1995 and 2000 we compare the health functionings (as indicated by life expectancy) that are achieved by the income growth generated by educational attainment, with the total health functionings that are achieved by educational attainment. We find that educational attainment (as indicated by average years of schooling) has a significant effect on life expectancy independently of its effect by way of income growth. A 1% increase in per capita income increases life expectancy by 0.073954% while a 1% increase in average years of schooling directly increases life expectancy by 0.055324%. Because it shows that income underestimates the health functionings achieved by educational attainment, our empirical findings lend support to the claim that the value of education should be measured in terms of the capability for functioning, rather than resources.