This paper explores the role of unobserved individual characteristics in the health-assets and health-portfolio correlations. We apply various econometrics models to a unique longitudinal dataset with rich information that allows for the exploitation of four different health indices. Our findings show strong cross-sectional correlations between health and both financial and non-financial assets, but these correlations seem to be mainly driven by heterogeneity as the correlations largely disappear in the fixed-effects model. Adverse health shocks, however, are found to motivate a safer portfolio choice even after individual fixed-effects are controlled for – a result consistent with the prediction made by the background risk theory. Our findings suggest that health shocks shift investment from risky assets toward other financial assets, but keep the total financial assets unchanged.