Antioxidant micronutrients have been hypothesized to provide protection against rheumatoid arthritis. We investigated serum selenium and serum alpha-tocopherol for their prediction of subsequent development of rheumatoid arthritis in a case-control study nested within a Finnish cohort of 18,709 adult men and women who had neither arthritis nor a history of it at the baseline examination in 1973-1978; by late 1989, 122 had developed rheumatoid arthritis. Of the incident cases, 34 were rheumatoid factor-negative. Three controls per each incident case were individually matched for sex, age, and municipality. Serum selenium and alpha-tocopherol concentrations were measured from stored serum samples collected at baseline. Serum selenium was inversely related to subsequent occurrence of rheumatoid factor-negative but not rheumatoid factor-positive rheumatoid arthritis. The relative risks, adjusted for smoking and serum total cholesterol, for the highest relative to the lowest tertile of serum selenium, were 0.16 [95% confidence interval (CI) = 0.04-0.69] for rheumatoid factor-negative and 0.96 (CI = 0.49-1.90) for rheumatoid factor-positive rheumatoid arthritis. During the first 10 years of follow-up, the relative risk for rheumatoid arthritis for the highest compared with the lowest tertile of serum alpha-tocopherol was 0.44 (CI = 0.19-0.99). No association was found for longer follow-up periods. Low selenium status may be a risk factor for rheumatoid factor-negative rheumatoid arthritis, and low alpha-tocopherol status may be a risk factor for rheumatoid arthritis independently of rheumatoid factor status.