Abstract Background We investigated prospectively among community-dwelling older adults aged 65 years and older whether a larger kyphosis angle is associated with poorer physical performance (balance, muscle strength, or both), and whether this association is unidirectional. Methods Male and female participants performed a multicomponent physical performance test with subscores for gait, muscle strength, and balance at baseline and after 2 years. Hand grip strength was also measured at baseline and at follow-up. The Cobb angle was measured on DXA-based Vertebral Fracture Assessments, made at the baseline and follow-up visit. Through linear and logistic regression analysis, we investigated the association between the kyphosis angle and physical performance and vice versa. We stratified for sex, and tested for effect modification by age and study center. Results The mean kyphosis angle was 37° and 15% of the participants (n = 1 220, mean age 72.9 ± 5.7 years) had hyperkyphosis (Cobb angle ≥50°). A larger kyphosis angle at baseline was independently associated with a poorer total physical performance score in women of the oldest quartile (≥77 years) in both the cross-sectional and longitudinal analyses (baseline B −0.32, 95% confidence interval [CI] −0.56–0.08; follow-up B 0.32, 95% CI −0.55–0.10). There was no association between physical performance at baseline and kyphosis progression. Conclusion A larger kyphosis angle is independently associated with a poorer physical performance at baseline and over time, and the direction of this association is unidirectional. These results emphasize the importance of early detection and treatment of hyperkyphosis to prevent further worsening of the kyphosis angle, thereby potentially preserving physical performance.