The cumulative incidence of rheumatoid arthritis (RA) was compared between different occupations, and between different exposure groups based on a job-exposure matrix (JEM). The study population comprised those subjects who in 1980 lived in one of 13 Swedish counties, were born between 1905 and 1945, and who had stated the same occupation in the censuses of 1960 and 1970, a total of 375,035 men and 140,139 women. The study population was followed concerning hospital care for rheumatoid arthritis in 1981-1983 by lineage to the Swedish Hospital Discharge Register. In general there were rather small differences in the relative risk of RA in different exposure groups and different occupations. Most of the occupations associated with an increased risk of RA were occupations in which it was possible to work when the disease was present, i.e. cost accountants, estimating clerks and working proprietors in the retail trade. However, an increased relative risk of RA was also observed in some occupations where selection of RA patients out of heavy work should have biased genuinely increased relative risks towards unity. Such occupations were farmers, upholsterers, lacquerers, concrete workers, and hair-dressers. Substantial handling of organic solvents, according to the JEM, was associated with an increased relative risk.