Mortality over 9 years in rheumatoid arthritis was studied according to baseline demographic, disease, therapy and comorbidity variables, and measures of functional capacity variables. Significant differences between patients who survived and died over the next 9 years were seen for 8 variables: age, joint count, oral corticosteroid use, presence of concurrent heart disease, formal educational level, and 3 quantitative measures of functional capacity, questionnaire responses regarding activities of daily living, modified walking time and the button test. Five-year survivals of 50% or less were seen in patients with severely dysfunctional values for the 3 quantitative measures of functional capacity. Increased relative risk of mortality according to functional capacity measures was not explained by age, sex, duration of disease, smoking history, joint count, hand radiograph score, grip strength, morning stiffness, formal educational level, oral corticosteroid or parenteral gold use, or various comorbidities, and was not expected by a majority of physicians.