The relation between full-body, radiographically defined osteoarthritis and survival was examined in a cohort of 296 women aged 42-76 years (mean age, 57.1 years). These women were a random sample of women with very low body burdens of radium who were part of a larger cohort of women first employed in the US radium dial-painting industry between 1915 and 1945. At entry into the study between 1957 and 1982, these women had a clinical examination, and full-body radiographs were taken. Fifty-five joints (18 joint groups) of the hands, feet, cervical spine, lumbar spine, pelvis, and knees in each woman were graded for osteoarthritis by the method of J. H. Kellgren and J. S. Lawrence (Ann Rheum Dis 1957; 16:494-502). Through 1985, 18.6% (n = 55) of the women died. Cox regression showed a decreased survival for women with an increasing number of joint groups affected with osteoarthritis after adjusting for age at examination (hazard ratio = 1.45 for each increase in 3.1 joint groups (1 standard deviation) affected with osteoarthritis, 95% confidence interval 1.12-1.87). Further adjustment for a history of diabetes, smoking, alcohol use, and body mass index only slightly altered the risk. Similar results were obtained for the number of joints with osteoarthritis and the number of structures (e.g., left hand and right hand) with osteoarthritis. These results suggest that an increasing prevalence of full-body radiographic osteoarthritis is associated with decreased survival independent of age and several comorbid conditions related to osteoarthritis.