The determinants of change in blood pressure during childhood were studied in 596 Dutch children, aged 5-19 years. The children were randomly selected from a general population in Zoetermeer, near the Hague, the Netherlands, and this report deals with 462 children (78%), who had three to seven annual measurements of blood pressure between 1975 and 1982. The rate of change in blood pressure was obtained by least-squares regression of blood pressure on time for each subject. Initial level of blood pressure was associated negatively with subsequent change in systolic and diastolic pressure, even after adjustment for regression toward the mean. Parental blood pressure and initial body weight were related positively, but weakly, to the rate of blood pressure change. Initial serum uric acid was associated positively with change in blood pressure in girls only. Vascular reactivity, as measured by the cold pressor test, was not related to the rate of change in blood pressure. The same applied to pulse rate and the use of tobacco, coffee, and oral contraceptives. The finding that children with the highest initial levels of blood pressure did not have the largest subsequent increase suggests that there is no "horse-racing" of blood pressure during childhood. There is some evidence that the relation between initial level and subsequent change in blood pressure is modified by age.