If the height of a prepubertal child is measured at one time and again a year later, any decision about the child's growth should be based on the distributions of the later height and height velocity conditional on the value of the initial height. The standard charts for height and height velocity derived by Tanner, Whitehouse and Takaishi (TWT) (1966) can be used in this context provided a correction is subsequently made, either by calculation or by reference to a suitable table. Moreover, the data provided by TWT, on which the charts are based, can be used to show that, over a certain age-range, the correlation between successive 12-month heights of a child is greater than 0.9 and approaches 1 with increasing age. On the other hand, the correlation between the height at any age and the velocity over the ensuing 12 months is low, while the correlation between two successive 12-month velocities is so low as to make it reasonable to treat them as independent. One phenomenon that has been much observed is that a child who grows along the third centile for height seems to grow along the 25th-30th centile for velocity. This effect follows when a conditional velocity is plotted onto a chart giving the distribution of an unconditional velocity. There is, moreover, a slight regression back to the population mean from any age to the next; this has little effect on the mean height but a noticeable effect on the mean velocity.