This paper empirically addresses two questions using a large, individual-level Swedish data set which links mortality data to health survey data. The first question is whether there is an effect of an individual's self-assessed health (SAH) on his subsequent survival probability and if this...