The effects of changing the percentile cut points of prior blood pressure measurements (both single and serial examinations) on prediction of subsequent high blood pressure, and of the definition of subsequent high blood pressure were assessed in 1,501 children from Bogalusa, Louisiana. Subjects were 2-14 years old at initial examination (year 1, 1973-1974) and were reexamined three, five, and eight years after the initial examination. Increasing the stringency of the prior measurement cut point resulted in increased specificity and positive predictive value, but decreased sensitivity and negative predictive value. For prediction of subsequent systolic blood pressure at the 90th percentile from year 1 levels, increasing the year 1 cut point from the 80th to the 95th percentile resulted in decreased sensitivity from 42.8 to 13.0%, and increased specificity from 83.0 to 97.3%. For systolic pressure, requiring all prior serial measurements to be above specified cut points resulted in further increases in specificity (maximizing at 100%) and decreases in sensitivity (minimizing at 2.1%). Using a year 1 cut point at the 90th percentile, increases in the definition of subsequent high blood pressure (from the 80th percentile to 140/90 mmHg or on treatment, respectively) resulted in increased sensitivity (from 20.5 to 33.3%) and negative predictive value, with decreasing specificity (from 93.9 to 91.3%) and positive predictive value. The results indicate that use of stringent criteria (serially at the 95th percentile) to identify children at risk for future essential hypertension will result in a substantial portion of the childhood population who will escape early identification but who will develop adult hypertension. Less stringent criteria will increase sensitivity and thereby provide the opportunity for primary prevention to a larger portion of the general childhood population who are at high risk for adult hypertension.