Objectives: to examine the association between 34-year trajectories of social activity, from middle age to old age and late-life disability.                  

Methods: data from the Swedish Level of Living Survey (LNU) and the Swedish Panel Study of the Oldest Old (SWEOLD) were used. LNU data from 1968, 1981, 1991 and 2000 were merged with SWEOLD data from 1992, 2002 and 2004 to create a longitudinal data set  with five observation periods. Trajectories of social activities covered 1968–2002, and late-life disability was measured  in 2004. The sample consisted of 729 individuals aged 33–61 at baseline (1968), who participated in at least four observation periods and who were free from mobility limitations at baseline. Four trajectories of social activity were identified and used as predictors of late-life disability.                  

Results: reporting low/medium levels of social activity from mid-life to old age was the most common trajectory group. Persons reporting continuously low/medium or decreasing levels of social activity had higher odds ratios for late-life disability (OR = 2.33 and OR = 2.15, respectively) compared with those having continuously high levels of activity, even when adjusting for age, sex and mobility limitations, and excluding persons with baseline mobility limitations.                  

Conclusions: results suggest that the disability risk associated with social activities is related to recent levels of activity, but also that risk may accumulate over time, as indicated by the higher disability risk associated with the continuously low/medium level social activity trajectory.