The prevalence of self-reported depressive symptoms was investigated in a case-control study of patients with rheumatoid arthritis (RA) attending an out-patient clinic at the Middlesex Hospital. Patients selected their own controls, matched for age and sex. Previous attempts to measure depressive symptoms in RA have suffered from measurement error due to criterion contamination, where psychological symptoms augment depressive scores. A total of 163 patients (77% of the sample) and 115 matched pairs completed the Hospital Anxiety and Depression Scale (HADS). The results indicated that RA patients are more depressed and anxious than controls. The prevalence of depression above the cut-point was 15%. This figure is comparable to other reports adjusted for criterion contamination, but is lower than that of other studies which employ 'contaminated' tools. The depression scale of the HADS appeared to be relatively free of criterion contamination. Subject to further reliability testing, the HADS may be a practical screening tool for practitioners to assess patients in need of psychological interventions.