Thirty-three patients diagnosed as suffering from definite or classic rheumatoid arthritis (RA) received conventional medical therapies and were randomly assigned to one of three adjunct treatment conditions. These were (a) thermal biofeedback training and cognitive-behavioral group therapy; (b) structured social support group therapy; and(c) no adjunct treatment. Subjects in both the cognitive-behavioral and the social support treatment conditions reported significant descreases in trait anxiety and depression from the pretreatment to posttreatment assessment. Only the cognitive-behavioral subjects showed significant decreases across assessment periods in pain behavior, self-reports of pain intensity, and rheumatoid factor titer. Subjects’perceptions of control over their RA symptoms were not associated with their outcomes. The results suggest that the cognitive-behavioral treatment producedpositive effects upon pain behavior and. immune system activity. The stability of these results will be examined using the entire sample of 53 subjects at six-month and one-year follow-up assessments. © 1985 Taylor & Francis Group, LLC.