Age-related changes in the posterior extensor muscles of the cervical and lumbar spine have been reported in some studies; however, longitudinal changes in the thoracic spine of healthy subjects are rarely reported. Therefore, this study aimed to evaluate changes in the cross-sectional areas (CSAs) of posterior extensor muscles in the thoracic spine over 10 years and identify related factors. The subjects of this study were 85 volunteers (mean age: 44.7 ± 11.5) and the average follow-up period was about 10 years. The CSAs of the transversospinalis muscles, erector spinae muscles, and total CSAs of the extensor muscles from T1/2 to T11/12 were measured on magnetic resonance imaging. The extent of muscle fat infiltration was assessed by the signal intensity (luminance) of the extensor muscles’ total cross-section compared to a section of pure muscle. We applied a Poisson regression model, which is included in the generalized linear model, and first examined the univariate (crude) association between each relevant factor (age, sex, body mass index, lifestyle, back pain, neck pain, neck stiffness, and intervertebral disc degeneration) and CSA changes. Then, we constructed a multivariate model, which included age, sex, and related factors in the univariate analysis. The mean CSAs of the transversospinalis muscles, erector spinae muscles, and total CSAs of the extensor muscles significantly increased over 10 years. Exercise habit was associated with increased CSAs of the erector spinae muscles and the total area of the extensor muscles. The cross-section mean luminance significantly increased from baseline, indicating a significant increase of fat infiltration in the posterior extensor muscles. Progression of disc degeneration was inversely associated with increased fat infiltration in the total extensor muscles.