Weight-for-height standards in children are usually constructed on the basis that the expected weight for a given height does not depend on age, an assumption which is unjustified. The present paper investigates regression standards of age-standardized weight for age-standardized height, the standardization being achieved by expressing weight and height as fractions of the 50th centile for age from a suitable growth standard. The precise choice of standard is not critical. Data on 4631 children from five different countries, exhibiting a wide spectrum of growth status, show that throughout childhood until puberty, the following ratio is appropriate as a simple and convenient index of weight-for-height: age-standardized weight/(age-standardized height)2. During puberty a larger power than 2 is required, so the index as specified is inappropriate. Approximate values for the distribution centiles of the index are suggested. The index may be used to assess degree of malnutrition or obesity, for individuals or groups seen on a single occasion. A slide-rule is described which calculates the index directly, given the child's sex, age, height and weight.