BACKGROUND AND PURPOSE:Treatments that prevent worsening kyphosis are important due to the progressive nature of kyphosis with aging. We assessed long-term efficacy of treatment effects after a short-term kyphosis exercise and posture training intervention in a cohort study among older adults with hyperkyphosis, and investigated whether long-term treatment effects differ among males and females. METHODS:In the original kyphosis intervention, 112 older adults enrolled in a waitlist design randomized controlled trial. One hundred three participants, mean age 70.0 (5.7) years and kyphosis 52.0° (7.4°), completed a twice weekly, 3-month, group exercise and posture training intervention, and were eligible to enroll in the follow-up study. We compared (1) change in outcomes pre-/postintervention to change postintervention over the follow-up period, (2) change in outcomes pre-/postintervention and postintervention to follow-up, stratified by sex, and (3) long-term change postintervention to follow-up in males and females. Primary outcome was change in kyphometer-measured thoracic kyphosis. Secondary outcomes were change in lumbar lordosis, objective measures of physical function, self-reported measures of physical activity, and health-related quality of life (HRQoL). RESULTS AND DISCUSSION:Forty-three participants, 42% of the eligible cohort, returned for follow-up, a mean 3.0 (0.7) years after completing the original intervention. Participants (27 females and 16 males) were 73.8 (6.1) years old, with mean kyphosis 48.9° (11.9°) at follow-up. Kyphosis declined -1.5° (95% confidence interval [CI]: -3.9° to 1.0°) postintervention to follow-up and this was no different than change pre-/postintervention, P = .173. Lordosis improved 8.9° (95% CI: 6.2° to 11.6°), more than change pre-/postintervention, P < .001. Gait speed measure of physical function increased 0.08 (95% CI: 0.02 to 0.14) m/s, Physical Activity Scale for the Elderly (PASE) measure of physical activity increased 4 (95% CI: -16 to 24) points, and Patient-Reported Outcomes Measurement Information System (PROMIS) mental health T-score measure of HRQoL increased 1.1 (95% CI: -1.0 to 3.1) points, but these improvements were not significantly more than change pre-/postintervention, P > .050. Other measures of physical function (modified Physical Performance Test [PPT], Timed Up and Go, and 6-minute walk) and HRQoL (Scoliosis Research Society [SRS-30] self-image and PROMIS physical function and physical health) declined at follow-up, significantly more than change pre/postintervention, P ≤ .050. Comparing change in outcomes pre-/postintervention and postintervention to follow-up, stratified by sex, both males and females increased lordosis, and decreased modified PPT and 6-minute walk measures of physical function, P < .050. Males and females differed in long-term change postintervention to follow-up. Time loaded standing and PASE improved in females compared with males, P = .008 and P = .092, respectively, and PROMIS mental health, physical health, and physical function declined in females compared with males, P = .073, P = .025, and P = .005, respectively. CONCLUSIONS:In our follow-up study, a mean of 3.0 (0.07) years after a 3-month kyphosis exercise and posture training intervention, kyphosis maintained and did not progress as expected with age. There was long-term improvement in lordosis. Compared with treatment effects from the short-term intervention, gait speed maintained equally well in males and females, while trunk endurance improved in females. Further investigation of long-term benefits of a short-term kyphosis exercise and posture training intervention is warranted.