Data on 2,006 patients with rheumatoid arthritis (RA) for 1984 and 1986 were analyzed from 5 American Rheumatism Association Medical Information Systems patient centers to assess covariates of future disability. The dependent variable was the Health Assessment Questionnaire disability index, measured in 1986. Independent variables, measured in 1984, include years of schooling, age, age squared, sex, labor force status, occupation, marital status, race, income, sedimentation rate, log latex, number of tender joints, and duration of illness. A negative association between schooling and the disability index is strongly apparent for men and weaker in women. Results for men persist even after adjustment for occupation and income, but not after additional adjustment for biological variables. The effects of schooling upon progression of RA are complex and interpretation requires simultaneous assessment of a variety of other variables. Causal effects of level of schooling are seen with a social science perspective which considers the biologic data to represent dependent variables yet cannot be inferred from a clinical model which considers the biology of the disease as an independent variable.