OBJECTIVE: Joint counts have a central role in assessing disease activity in rheumatoid arthritis (RA). They are usually undertaken by physicians or nurses. We investigated whether joint counts can be devolved to patients and evaluated the use of a patient-derived Disease Activity Score (DAS). METHODS: One hundred RA patients attending a specialist unit were evaluated, comparing joint counts by a physician with patient-assessed joint counts and DAS derived from both methods. They were related to other measures of disease activity in the European League Against Rheumatism (EULAR) core data set and with the Rheumatoid Arthritis Disease Activity Index (RADAI; a validated patient self-assessment index). RESULTS: Regression analysis showed no significant differences between a physician's and patient's joint counts and DAS. There were middle to high correlations between patient and physician assessments of tender joints and swollen joints; using R2, this explained 70% of the variance for tender joints and 40% for swollen joints. Kappa analysis showed good agreement between physician and patient assessments of individual joint tenderness (kappa values 0.49-0.84). There was lower agreement for individual swollen joints (kappa values 0.02-0.61). Physician DAS and patient DAS had a similar correlation with the Health Assessment Questionnaire (HAQ) (r = 0.50 and r = 0.48, respectively). CONCLUSION: The agreements between physician and patient assessments are sufficient to allow patients' assessments to be used for clinical research. This is especially the case with a patient-derived DAS. However, the results are not directly interchangeable and further studies are needed before patients' assessments are used to guide clinical practice.