The prevalence of scoliosis is not known in patients with idiopathic short stature, and the impact of treatment with recombinant human growth hormone on those with scoliosis remains controversial. We investigated the prevalence of scoliosis radiologically in children with idiopathic short stature, and the impact of treatment with growth hormone in a cross-sectional and retrospective cohort study. A total of 2,053 children with idiopathic short stature and 4,106 age- and sex-matched (1:2) children without short stature with available whole-spine radiographs were enrolled in the cross-sectional study. Among them, 1,056 with idiopathic short stature and 790 controls who had radiographs more than twice were recruited to assess the development and progression of scoliosis, and the need for bracing and surgery. In the cross-sectional study, there was an unexpectedly higher prevalence of scoliosis (33.1% (681/2,053) vs 8.52% (350/4,106)) in children with idiopathic short stature compared with controls (odds ratio 3.722; p < 0.001), although most cases were mild. In the longitudinal study, children with idiopathic short stature had a higher risk of the development and progression of scoliosis than the controls. Among children with idiopathic short stature without scoliosis at baseline, treatment with growth hormone significantly increased the risk of developing scoliosis (p = 0.015) and the need for bracing (p < 0.001). Among those with idiopathic short stature and scoliosis at baseline, treatment with growth hormone did not increase the risk of progression of the scoliosis, the need for bracing, or surgery. The impact of treatment with growth hormone on scoliosis in children with idiopathic short stature was considered controllable. However, physicians should pay close attention to the assessment of spinal curves in these children.