Major musculoskeletal conditions including arthritis represent an increasing burden on individuals and societies. We analyzed the association between self-reported arthritis and mortality in the U.S. elderly disabled and non-disabled individuals using unique disability-focused data from the large-scale population-based National Long Term Care Survey. It was found that males and females who reported arthritis/rheumatism have, generally, smaller risks of death than those who did not report those conditions. This inverse relationship is more pronounced in disabled individuals. This finding holds for both short-term (relative risk [RR] = 0.81; 95% confidence interval [CI] = 0.75-0.88 for males and RR = 0.76; CI = 0.71-0.82 for females) and long-term follow-ups (RR = 0.82; CI = 0.78-0.87 for males and RR = 0.83; CI = 0.79-0.87 for females). For females, this effect is age insensitive, while for males it is limited to ages below 85. Demographic and 19 major self-reported geriatric conditions have trivial effect on these risks, supporting the view that a better survival of diseased individuals can be attributed to the effects of medical treatment. Given the widespread prevalence of arthritis/rheumatism and disability in elderly populations and the increasing population of the elderly, these findings call for comprehensive analyses of factors driving better survival and medical costs associated with extended lives.