Background: Falls are a common cause of morbidity among older adults. Although many fall risk factors are well understood, it is unknown whether a common geriatric condition, hyperkyphosis, increases fall risk. Objectives: This dissertation research sought to determine whether older persons with worse degrees of kyphosis may be at increased risk of falls. Methods: Three longitudinal cohorts of older adults (aged ≥65) were used, each with different kyphosis measures, to examine the association between degree of kyphosis and fall risk. Study 1 examined whether four measures of kyphosis (Debrunner kyphometer, flexicurve ruler, Cobb angle, and block method) increased the odds of an incident fall over one year of monthly falls follow-up in a mixed-gender sample of 72 individuals (mean age=78). Studies 2 and 3 each prospectively examined two different measures of kyphosis and incident falls over three years of tri-annual follow-up through multivariable modeling with GEEs. Study 2 included 2,346 men with the Cobb angle measure of kyphosis (mean age=74) and 2,929 men with block-measured kyphosis (mean age=79) from the Osteoporotic Fractures in Men Study (MrOS). Study 3 used data from 1,162 women with the Cobb angle measure (mean age=69) and 610 women with the flexicurve measure of kyphosis (mean age=73) from the Study of Osteoporotic Fractures in Women (SOF). Results: In Study 1, all four measures of kyphosis conferred a 2-3 fold increase per SD in the odds of a fall over the following year (64% of participants fell over one year). In Studies 2 and 3, Cobb angle was not associated with fall risk in either MrOS or SOF. The blocks method was predictive of fall risk in MrOS (RR=1.11 per SD; 95% CI=1.06, 1.17) and KI was predictive of fall risk in SOF (RR=1.22 per SD; 95% CI=1.07, 1.40). Conclusion: All three studies provide evidence that increased kyphosis predicts greater fall risk in older persons. The Cobb angle measure, however, appears to be least related to falls, possibly because it is a measure limited to the mid to lower thoracic spine whereas the block and flexicurve methods include curvature in the lower cervical and entire thoracic spine.