In order to explore the concept that scoliosis is fundamentally a loss of left-right symmetry. surface topography was used to measure asymmetry in three dimensions at three levels on the back surface. Statistical analysis of prospectively collected topographic, radiographic and clinical data, in girls with adolescent idiopathic scoliosis, was carried out and comparisons were made with theoretically perfect symmetry (test value of zero). All scoliosis showed statistically significant differences in coronal dimensions, index points on the convex side of the scoliosis being further from the mid-line than those on the concave side. Primary thoracic scoliosis differed from thoracolumbar and lumbar in that they showed directional asymmetry at all levels and in all directions, the side of the scoliosis convexity being broader, taller and thicker. This asymmetry is not due to posture, spinal balance or trunk rotation, as left and right sides are being compared independently of their orientation in space. The asymmetry is of size in three dimensions and size is determined by growth. Growth is a three dimensional process, but does not necessarily occur equally in all three. Differential growth is both directional and regional, particularly during the pubertal growth spurt, when proportions change substantially, and is controlled by many genes, as well as by hormones and signalling molecules. The implication is that scoliotic deformity is the result of asymmetric growth, not confined to the vertebrae, but affecting the entire trunk. This is a developmental, rather than pathological, phenomenon. It makes questions of aetiology redundant and natural history logical.