© 2017 The Author(s). Background: Hyperkyphosis, an excessive anterior curvature in the thoracic spine, is associated with reduced health status in older adults. Hyperkyphosis is highly prevalent, more common in older women than men. There is no standard intervention to reduce age-related hyperkyphosis. Sex differences in response to a kyphosis-specific exercise intervention are not known. Methods: We conducted a randomized controlled trial of a targeted kyphosis-specific exercise and postural training program on the primary outcome Cobb angle of kyphosis, and investigated whether the magnitude of change differed between men and women. One hundred twelve participants aged ≥60 years with kyphosis ≥40° were enrolled and randomized to exercise or waitlist control, and 101 participants had analyzable baseline and follow-up radiographs for Cobb angle measurements. A group intervention including 10 participants per group was delivered by a physical therapist, 1-h, twice a week for 3-months. Controls were placed on a waitlist for 3 months before receiving a delayed intervention. Primary outcome was change from baseline to 3-months in Cobb angle measured from standing lateral spine radiographs. Secondary outcomes included change over 3-months in kyphometer-measured kyphosis, physical function and quality of life. Groups were combined for analysis after both received the intervention, and sex differences in response to the intervention were tested with ANOVA. Results: Participants (60 women, 41 men) were 70.0 (SD = 5.7) years old with mean Cobb angle 55.9 (SD = 12.2) degrees at baseline. The active group had higher baseline modified Physical Performance Test scores than control, p = 0.03. Men had greater baseline kyphometer-measured kyphosis, p = 0.09, and higher bone mineral density (BMD), spine strength, more vertebral fractures and diffuse idiopathic skeletal hyperostosis (DISH) than women, p ≤ 0.01. There was no statistically significant difference between groups in change in Cobb at 3-months, p = 0.09, however change in kyphometer-measured kyphosis differed by 4.8 (95% CI:-6.8,-2.7) degrees, p < 0.001, favoring the active group. There were no differences between men and women in change in either kyphosis measurement after intervention, p > 0.1. Conclusions: A 3-month targeted spine strengthening exercise and posture training program reduced kyphometer-measured, but not radiographic-measured kyphosis. Despite sex differences in baseline kyphosis, BMD, spine strength, fractures and DISH, sex did not affect treatment response. Trial registration: ClinicalTrials.gov Identifier: NCT01766674.