Reliable, standardized balance tests for dogs are not available yet. The purpose of this study was to investigate the reliability of static and dynamic posturography in healthy dogs. Healthy dogs (n = 20) were positioned with four paws longitudinally and with the forepaws only transversely on a modified pressure-sensitive balance platform (Posturomed-FDM-JS, Zebris, Isny, Germany). Three static and dynamic posturographic trials were recorded (recording duration: 20 s) and repeated after 7–14 days. Center of pressure (COP) parameters COP-path-length (PL; mm), 95% COP-confidence-ellipse-area (CEA; mm2), and COP-average-velocity (AV; mm/s) were calculated for the first steady-state 5 s intervals of each trial. The reliability of COP parameters was assessed with robust linear mixed effects models with nested random effects of patient and trial. The training effect was analyzed using Cohen’s d. For static posturography, PL, CEA, and AV did not differ significantly between time points; CEA had the highest reliability (p = 0.92). For dynamic posturography, AV and PL differed significantly between time points (AV: p ≤ 0.043; PL: p ≤ 0.045). Slight training effects were observed for transverse positioning (Cohen’s d: PL 0.65; AV 0.267) and moderate training effects for longitudinal positioning (Cohen’s d: PL: 0.772; AV: 0.783). This study showed that static posturography on a modified Posturomed-balance platform was reliable in healthy dogs but indicated training effects during dynamic posturography.