Summary Several studies have reported changes in body composition in rheumatoid arthritis (RA) and ankylosing spondylitis (AS). Our study showed that body composition measurements obtained by absorptiometry were highly reproducible in patients suffering from these diseases. This study justifies the use of absorptiometry measurements in longitudinal studies in this population. Purpose Our study aimed to assess the reproducibility of total and regional body composition in patients with rheumatoid arthritis (RA) and with ankylosing spondylitis (AS) and to compare them to healthy subjects. Methods The study enrolled 80 subjects including 32 healthy subjects, 31 RA patients, and 17 AS patients. Each subject had two scans in one day under the same standard conditions and none ate nor drunk before being repositioned on the table. The reproducibility was assessed through the coefficient of variation (CV), the least significant change (LSC), the intraclass correlation (ICC), and the smallest significant difference (SDD). Results Total body composition measurements obtained by dual-energy X-ray absorptiometry (DXA) were highly reproducible, and there was no statistically significant difference between reproducibility in healthy subjects, patients with RA, and patients with AS. For total body fat mass (FM), lean mass (LM), and bone mineral content (BMC) in the total population, CV values were 1.71%, 1.25%, and 1.74%, respectively; ICC values were 0.998, 0.996, and 0.993, respectively; LSC values were 4.88%, 3.7%, and 5.2%, respectively; and SDD values were ± 1.23 Kg, ± 1.47 Kg, and ± 126.0 g, respectively. For regional body FM, LM, and BMC in the total population, CV values in the arms were 8.46%, 4.17%, and 3.79%, respectively; in the legs 6.24%, 3.59%, and 2.04%, respectively, and in the trunk 5.02%, 2.92%, and 5.24%, respectively. Conclusion Total body tissue mass, FM percentage, FM, LM, and BMC measurements obtained by DXA are highly reproducible in RA and AS.