We sought to assess occupational joint use and osteoarthritis (OA) longitudinally in a large population with multiple occupations. Subjects were members of the Framingham Heart Study cohort followed over 40 years with occupational status assessed at the beginning of the Heart Study [from Examination 1 (1948-51) through Examination 6 (1958-61)] and knee OA assessed by weight bearing knee radiograph at Examination 18 (1983-85) when mean age of subjects was 73 years. Each subject's job was characterized by its level of physical demand and whether the job was associated with knee bending. Odds ratios (OR) testing the association of job demand with OA were adjusted by logistic regression for age, body mass, knee injury history, smoking, and educational level. Men whose jobs required knee bending and at least medium physical demands had higher rates of later radiographic knee OA (at least definite osteophytes) than men whose jobs required neither (43.4 vs 26.8%; OR of OA = 2.22, 95% CI 1.38, 3.58). Rates of severe radiographic OA (osteophytes and joint space narrowing) and of bilateral radiographic OA were also significantly increased in these men. Few women had jobs requiring knee bending or that were physically demanding and these jobs were generally unassociated with later radiographic OA. Only a small number of men (n = 28) had symptomatic knee OA, and we could not confirm that it was associated with occupation in men. Thus, among men, occupations which combine knee bending and physical demands may be an important cause of radiographic OA.