This paper reports a new method for expressing numerically asymmetry of the contour of the back in a forward-bending position. Information is given at three spinal levels (T8, T12 and L3) for 636 schoolchildren aged 8 to 15 years. Rib-hump and lumbar-hump scores were standardised to create trunk asymmetry scores (TASs) making comparison possible between children of different age, size and sex. Two groups of children were defined: those with clinically straight spines (585 children); and those with clinical evidence of lateral spinal curves (51 children). In the children with clinically straight spines the main findings were: about 1:4 had objectively detectable rib and lumbar humps; female-to-male ratios were 1.2:1 for the thoracic region and 1.4:1 for the lumbar region; right humps were about 10 times more common than left; TASs in the boys and girls at each spinal level had normal distributions about means to the right of zero (where zero represents perfect symmetry); at T8 and T12, a wider scatter of TASs in girls than in boys; at L3, larger TASs in girls than in boys; a relation between shortening of one lower limb and a contralateral hump on the back; and no relation to age (except at L3), stature (corrected for age) or handedness. The findings are discussed in relation to possible causes of back contour asymmetry, early diagnosis of scoliosis by screening, sexual dimorphism and significance for the pathogenesis of idiopathic scoliosis. Ten children with clinically straight spines and larger TASs, and 42 out of 51 children with clinical evidence of lateral spinal curves in the forward-bending position attended for radiographic examination. Twelve children had "scoliosis curves" of 11 degrees or more as defined by the Scoliosis Research Society. The results are reported in relation to TASs, spinal curve angle (Cobb) and vertebral rotation.