OBJECTIVE: To explore the optimal frequency of whole-body vibration training for improving the balance and physical performance in older people with chronic stroke. DESIGN: a single-blind randomized controlled trial. SETTING: Two rehabilitation units in the Wuhan Brain Hospital in China. PARTICIPANTS: A total of 78 seniors with chronic stroke. INTERVENTIONS: Low-frequency group (13 Hz), high-frequency group (26 Hz), and zero-frequency group (Standing on the vibration platform with 0 Hz) for 10 sessions of side-alternating WBV training. MAIN MEASURES: The timed-up-and-go test, five-repetition sit-to-stand test, 10-metre walking test, and Berg balance scale were assessed pre- and post-intervention. RESULTS: Significant time × group interaction effects in five-repetition sit-to-stand test (p = 0.014) and timed-up-and-go test at self-preferred speed (p = 0.028) were observed. The high-frequency group outperformed the zero-frequency group in both five-repetition sit-to-stand test (p = 0.039) and timed-up-and-go test at self-preferred speed (p = 0.024) after 10-sessions training. The low-frequency group displayed only a significant improvement in five-repetition sit-to-stand test after training (p = 0.028). No significant within- or between-group changes were observed in the Berg balance scale and walking speed (p > 0.05). No significant group-difference were found between low-frequency and high-frequency groups. No adverse events were reported during study. CONCLUSIONS: Compared with 13 Hz, 26 Hz had no more benefits on balance and physical performance in older people with chronic stroke.