Despite application of ultrasound for quantitative measurement of spinal curvatures has been reported with various studies, a systematic review for such is lacking. This systematic review aimed to evaluate (1) reliability of ultrasound; (2) validity of ultrasound using radiographic measurement as gold standard in idiopathic scoliosis patients; and (3) the use of various anatomical landmarks for measurement of spinal curvatures. MEDLINE, EMBASE, CINAHL, and CENTRAL databases were searched. QUADAS-2 quality assessment tool was adopted. Reliability of ultrasound in terms of intra-class correlation coefficient was recorded. Pearson correlation coefficients between ultrasound and radiographic measurements were extracted for meta-analysis. Subgroup analyses based on ultrasound measurement protocols of spinous process (SP), transverse processes (TP) and center of lamina (COL) were conducted. Eleven articles reporting 18 correlation analyses on 766 subjects were eligible for meta-analysis. The mean inter-rater reliability of ultrasound measurement was 0.87±0.07. Pooled correlation for all studies was 0.918 (95% CI: 0.868–0.949), exhibiting substantial heterogeneity (I²=90.50%, p<0.001). Subgroup analyses showed that pooled correlations were 0.887 for COL method (comprising 356 subjects); 0.924 for SP method (255 subjects); and 0.941 for TP method (117 subjects); all with notable heterogeneity (I²>90%, p<0.001). The overall risk of bias was rated moderate; yet publication bias was noted. Evidences showed that ultrasound was a promising non-invasive method with satisfactory validity and reliability for measuring coronal curvatures utilizing the SP, TP or COL methods. Further development of three-dimensional ultrasound towards scoliosis assessment will facilitate its translational application for managing scoliosis.