The perception of verticality is formed through the integration of multisensory gravitational information, including somatosensory, visual, and vestibular inputs. Older adults exhibit visual dependency (VD) as they rely more on visual information to compensate for reduced somatosensory and vertical perception. Increased VD is associated with falls, and the dynamic subjective visual vertical (SVV) is used to assess VD. However, the measurement reliability of dynamic SVV and VD using virtual reality (VR), which has garnered considerable attention in recent years, remains unclear. Therefore, our purpose in this study was to assess the test-retest reliability of dynamic SVVs and VDs using a VR method. We evaluated static and dynamic SVV of 40 older adults using a smartphone-based VR system (SVR-SVV). Dynamic SVV consisted of numerous spheres on the background rotating clockwise (CW-SVV) or counterclockwise (CCW-SVV). Each SVV measurement consisted of one set of 10 trials. VD was calculated as the mean value of dynamic SVV minus the mean values of static SVV. A re-test was conducted after one week. Reliability was analyzed using Bland-Altman plots and intraclass correlation coefficients (ICC 2, k) for each SVV measure. We observed no systematic bias in any of the SVV values, which were 0.1° (SD = 2.8°), 13.3° (SD = 8.3°), -12.8° (SD = 6.9°), and 15.7° (SD = 8.1°) for static SVV, CW-SVV, CCW-SVV, and VD, respectively. Test-retest reliability was good for static SVV (ICC = .817, p < .001), CW-SVV (ICC = .896, p < .001) and excellent for CCW-SVV (ICC = .914, p < .001), VD (ICC = .937, p < .001). The dynamic SVV and VD measurements using SVR-SVV demonstrated good test-retest reliability. Moreover, the SVR-SVV is more portable than conventional methods, making it highly useful in clinical practice.