Objectives: To assess the visual influence through photoelastic sole image on upright posture in hemiplegic patients.Design: Before-during trial.Setting: An academically based rehabilitation setting.Patients: Nineteen hemiplegic patients 3 to 8 months after the stroke were designated as severe (n = 5), moderate I (n = 4), moderate II (n = 5), and mild groups (n = 5). The severe, moderate, and mild groups were in Brunnstrom's motor recovery stage III, IV, and V, respectively. Vibration sense in the paretic leg was intact in the moderate II and mild groups, but impaired in the severe and moderate I groups.Intervention: Visual feedback therapy on standing before and while watching his or her own photoelastic sole image on a television screen (before vs during).Main Outcome Measure: The body-weight-bearing ratio on the paretic sole was analyzed.Results: The body-weight-bearing ratio on the paretic sole increased during visual feedback (before vs during, percentage, severe, 21.8 +/- 11.4 vs 38.7 +/- 13.7, p < .02; moderate I, 30.4 +/- 11.1 vs 39.3 +/- 6.4, not significant; moderate II, 32.8 +/- 10.3 vs 46.8 +/- 7.8, p < .05; mild, 29.7 +/- 7.0 vs 4.8 +/- 7.5, p < .005; mean +/- SD).Conclusion: As the only difference between the moderate I and II groups was impaired or intact vibration sense in the paretic limbs, vibration sense along with motor recovery is an important factor to increase body weight loading on the paretic sole in hemiplegic patients.