This paper illustrates a health index that uses multiple correspondence analysis to reduce multiple discrete indicators to a continuous variable using minimal modeling assumptions. The continuous index enables fixed effects estimators when health is the dependent variable, and it exhibits more variation than self-assessed health (SAH) to support identification when health is an independent variable. I compute this index for 18 waves of the British Household Panel Survey, describe its features, and compare its use to SAH. The index performs better in dynamic models of health and subsumes the explanatory power of SAH in wage equations.