ABSTRACT.Purpose: Retinal oximetry (RO) has been established as a non-invasive method to analyse oxygen saturation in retinal vessels. The aim of our study was to determine the reproducibility of RO images in healthy and in diseased retinas. Methods: A total of 61 right eyes (244 RO images) in 61 subjects (35$, 26#) were examined: 22 controls, 18 patients with glaucoma and 21 patients with inherited retinal diseases (IRDs). Four test-retest RO images were obtained in each subject. Oxygen saturation was measured with the oxygen saturation measurement tool of the Retinal Vessel Analyser (RVA; IMEDOS Systems UG, Jena, Germany). The test-retest standard deviation within the subject's (AESD w ) measurements (the mean vessel oxygen saturation in retinal venules and arterioles), its coefficient of variation (CoV) and the intraclass correlation coefficients (ICC) were analysed. Results: The average test-retest SD w in venules was AE2.52% (CoV = 4.35%) and in arterioles was AE1.67% (CoV = 1.76%). Among controls, glaucoma eyes and eyes with IRDs, the test-retest SD w in venules were AE2.33% (CoV = 4.48%), AE2.85% (CoV = 4.71%) and AE2.43% (CoV = 3.90%) (SD w p = 0.366 (CoV p = 0.452); one-way ANOVA). The test-retest SD w in arterioles were AE1.65% (CoV = 1.80%), AE1.83% (CoV = 1.92%) and AE1.54% (CoV = 1.56%), respectively [SD w p = 0.762 (CoV p = 0.686)]. The ICCs in venules were 0.76 in controls, 0.69 in patients with glaucoma and 0.82 in patients with IRD. The ICCs in arterioles were, respectively, 0.92, 0.70 and 0.93. Conclusion: The reproducibility of RO in healthy, as well as in diseased retinas, is excellent. In the glaucoma group, the lower standard deviation between subjects (SD b ) for arterioles contributes to the lower ICCs. Nevertheless, the measurements of oxygen saturation in arterioles seem more reliable when compared to venules.