If patients with rheumatoid arthritis (RA) could self-report symptoms in a manner which correlated with laboratory measures of inflammation this would be a valuable research or clinical tool. Developing such a tool means ensuring that the questions are understood and a variety of combinations of signs, symptoms and joints explored. Preliminary studies in two groups of 20 RA out-patients established an acceptable self-report questionnaire format by which patients could identify their joints. Fifty RA out-patients completed these self-report forms on four grades of each of four symptoms (pain, heat, stiffness, swelling) in each of 64 joints, as well as visual analogue scales (VAS) on overall pain and perceived disease activity. A clinical research assistant recorded the Thompson-Kirwan articular index (TKAI) and plasma viscosity (PV) was measured. The data were analysed in a variety of ways in an attempt to construct a self-report articular index (SRAI) which correlated with PV. The strongest models were then tested in 11 in-patients undergoing a flare of their disease. No adequate SRAI could be constructed which correlated with PV and in addition neither VAS score correlated with PV. There was a moderate correlation between the TKAI and a patient SRAI using the same joints, symptoms and weightings (r = 0.6, P < 0.01). Patients can clearly report different grades of multiple symptoms in multiple joints, but such reports cannot be shown to be a reliable indicator of inflammatory activity.