The three commonly used methods of height prediction employ various combinations of anthropometric data and bone age. In normal children, the regression type methods are preferable, though they do not perform well in the more severe disorders of growth. They are extremely dependent upon the limitations of the method of bone age determination that is used. A major source of error is the inability to predict the timing or the intensity of the adolescent growth spurt. Until this proves possible it is unlikely that significant improvements can be made.