The aim of this study was to investigate the reliability and validity of the self-report Thompson articular index (ThAI) in Dutch patients with rheumatoid arthritis (RA). A rheumatologist assessed the ThAI in 43 patients with RA. Patients completed the self-report ThAI and the AIMS-2 questionnaire to assess physical function, pain, mood and level of tension. Blood samples were taken to measure the erythrocyte sedimentation rate (ESR). After 4 weeks, patients were sent a questionnaire for a repeat assessment of the self-report AI. The test-retest reliability of the self-report ThAI was adequate (ICC=0.83). There was low agreement between ThAI scores from patients and AI scores assessed by the rheumatologist (ICC=0.44). Self-report ThAI scores (mean=230.5) were significantly higher than the rheumatologist's scores (mean=110.8). Levels of agreement between patients and rheumatologist for individual joints were disappointing, ranging from 49% to 74% (Cohen's kappa from -0.02 to 0.48). The rheumatologist's ThAI scores correlated significantly with ESR (r=0.55) and physical function (r=0.44), but not with pain, mood or level of tension. Patients' scores correlated significantly with physical function (r=0.51), pain (r=0.43), and mood (r=0.36) but not with ESR or level of tension. In regression analyses the only significant predictor of the rheumatologist's ThAI scores was ESR, and for patients' scores physical function, thus showing that patients' responses are not confounded by mood or level of tension. In conclusion, the self-report ThAI is a reliable measure, but the validity is questionable because of the non-significant correlation with ESR and the low level of agreement between patients and rheumatologist. The results indicate that self-reported joint involvement is more closely related to physical function than to arthritic activity.