The Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), the Bath Ankylosing Spondylitis Function Index (BASFI) and the Bath Ankylosing Spondylitis Global Score (BAS-G) (ranges 0-10) have gained widespread in use as self-reported measures of disease activity, functional impairment and overall well-being in patients with ankylosing spondylitis and other spondyloarthropathies (SpA). In Denmark, BASDAI, BASFI and BAS-G are systematically used to monitor treatment response in patients treated with tumour necrosis factor (TNF) inhibitors. The purpose of the present study was to examine the reproducibility of the indices in anti-TNF-treated SpA patients already familiar with the use of the indices. Testing was performed twice on two different days (median interval 7 days, range 4-10 days) under standardised conditions in 26 out-clinic patients (median age 39 years, range 22-56 years). Limits of agreement were calculated as the 95% likely range for the difference between paired scores. Test-retest results were significantly intercorrelated with r (s) = 0.90 for BASDAI, 0.92 for BASFI and 0.74 for BAS-G. Limits of agreement for BASDAI, BASFI and BAS-G were +/-1.8, +/-1.4 and +/-3.2, respectively. Reproducibility as expressed as the mean of individual standard deviations was significantly poorer for BAS-G than for BASDAI and BASFI (p < 0.01). Internal consistency reliability and construct validity of BASDAI and BASFI were acceptable. In conclusion, in a sample of anti-TNF-treated patients experienced with the use of BASDAI, BASFI and BAS-G, random measurement errors of the scores were not negligible. The finding should be considered when monitoring anti-TNF treatment in daily clinical practice.