Composite disease activity scores are frequently used in daily practice as tools for treatment decisions in patients with rheumatoid arthritis (RA). If reliable, patient-reported disease activity may be time saving in the busy clinic. The objective was to examine the test–retest reliability of the Disease Activity Score 28 CRP (DAS28-CRP) with four variables (4v) and three variables (3v), the Simplified Disease Activity Index (SDAI) and the Clinical Disease Activity Index (CDAI) when based on patient self-assessment of tender and swollen joints and to examine the agreement between these scores and physician-derived scores. Thirty out-clinic RA patients with stable disease were included. A joint count was performed two times 1 week apart by the patient and by an experienced physician. Test–retest reliability was expressed as the least significant difference (LSD), as the LSD in percent of the mean score (%LSD) and as intra-individual coefficients of variation (CVi). Mean scores based on physician vs. patient joint counts (visit 1) were: DAS28-CRP(4v) 3.5 ± 1.0 vs. 3.6 ± 1.1 (not significant (NS)), DAS28-CRP(3v) 3.4 ± 0.9 vs. 3.5 ± 0.9 (NS), SDAI 14.2 ± 9.4 vs.14.1 ± 9.4 (NS) and CDAI 13.4 ± 9.3 vs. 13.3 ± 9.4 (NS). The LSDs (%LSD) for duplicate assessments of patient-derived scores (visit 2 vs. 1) were: DAS28-CRP(4v) 0.8 (23.2), DAS28-CRP(3v) 0.9 (25.2), SDAI 8.3 (59.9) and CDAI 8.4 (63.8). Similar LSDs were found for differences between duplicate assessments of physician-derived scores and for differences between physician and patient-derived scores. CVis for SDAI and CDAI were significantly higher than for DAS28-CRP(4v) and DAS28-CRP(3v) (p < 0.005). Patient- and physician-derived scores agreed closely on group level. On the individual level, the LSDs between patient- and physician-derived scores were considerable but corresponded to both patient and physician intra-observer LSDs. Thus, scores based on patient-performed joint counts may be an alternative to traditional physician-derived scores in patients with stable disease.