Background: Hyperkyphosis may cause balance impairment in elderly people. Although the effectiveness of orthoses for improving balance in hyperkyphotic elderly people has received much attention, the mechanisms by which devices affect balance remain unknown. Objectives: The objective of this study was to evaluate changes in balance performance, thoracic kyphosis angle, craniovertebral angle, back muscle endurance and joint position sense after 3 months of wearing a Spinomed orthosis. The study also included a secondary exploratory analysis to determine whether changes in any of the above-mentioned outcome measures can predict balance performance improvement in elderly people with hyperkyphosis. Study design: Parallel group randomized controlled trial. Methods: In total, 44 hyperkyphotic elderly people were randomly allocated to an experimental group, who wore a Spinomed orthosis and a control group, who did not. No other treatment or change in physical activity was permitted during the study. A blinded assessor evaluated thoracic kyphosis angle, joint position sense, craniovertebral angle, back muscle endurance, Timed Up and Go Test time and Berg Balance Scale score at baseline and after 5, 9 and 13 weeks. All dependent variables were measured without the orthosis and analyzed separately using a 2 × 4 (time × group) mixed model analysis of variance. Based on the results of correlation analysis, thoracic kyphosis angle, back muscle endurance and joint position sense were selected as independent variables in a stepwise multiple regression model. Results: The two-way (group × time) interactions were significant in terms of Berg Balance Scale ( F = 11.6, P ⩽ 0.001, [Formula: see text]), Timed Up and Go Test ( F = 3.74, P = 0.013, [Formula: see text]), thoracic kyphosis angle ( F = 43.39, P ⩽ 0.001, [Formula: see text]), craniovertebral angle ( F = 5.245, P = 0.002, [Formula: see text]) and joint position sense ( F = 4.44, P = 0.005, [Formula: see text]). The two-way interaction was not significant in terms of back muscle endurance; however, the main effect of group was significant for this variable ( F = 3.85, P = 0.025). Stepwise multiple regression showed that thoracic kyphosis angle and joint position sense were significant determinants of Timed Up and Go Test time ( R2 = 0.155, P = 0.037 and R2 = 0.292, P = 0.012, respectively). Conclusion: Wearing a Spinomed orthosis for 3 months improved the posture, position sense and muscle performance of hyperkyphotic elderly people. Orthoses may improve balance performance by correcting spinal alignment and increasing proprioceptive information.