Objectives. Self-reported health has a well established relationship to later mortality, although the reasons are not entirely understood. This study examined the association of a similar self-reported measure of fitness with mortality and compared it to that of self-reported health. Design. The study had a prospective cohort design with multiple sampling points. Methods. Participants were 858 men and women from Glasgow and the surrounding area of Scotland, aged 59 when self-reported health and fitness data were first collected in 1991/92. They were re-interviewed at age 64 and 69 and mortality was tracked for 16.5 years in total. Hazard ratios for all cause mortality were estimated for those that reported poor health or poor fitness relative to others their age, taking into account a range of covariates, some of which were also time varying. Results. In both unadjusted and covariate adjusted models, self-reported fitness was at least as good a predictor of mortality as self-reported health. In a mutually adjusted model, both again emerged as significant predictors. Poor subjective health with poor subjective fitness appeared to be a particularly lethal combination. Conclusion. Both self-reported health and self-reported fitness were independent predictors of mortality. Where the objective assessment of aerobic fitness is not feasible, a simple measure of subjective fitness could prove a useful alternative.