OBJECTIVE The objective of this study was to examine the reproducibility of K4 and K5 diastolic blood pressure measurements and the ability to predict adulthood values. METHODS The Bogalusa Heart Study is a long-term epidemiologic study of cardiovascular disease risk factors from birth to early adulthood conducted in the biracial (one third black, two thirds white) community of Bogalusa, Louisiana. Analyses included blood pressure measurements taken on 12,139 subjects during multiple cross-sectional screenings from 1973 to 1994; 20% (N = 2530) had measurements taken as a child, ages 4 to 18 years, and during adulthood, ages 19 to 32 years. Six resting blood pressure measurements were taken by trained observers with mercury sphygmomanometers at each screening with K1, K4, and K5 recorded. Variance components analysis was used to evaluate the reliability of K4 and K5. RESULTS The total variance was larger for K5 (253 mm Hg2) than for K4 (109 mm Hg2) at age 5 years. Variance for both K4 and K5 decreased with age. The interobserver variability was larger for K5 (more than 50% vs 40%). Childhood K4 (vs childhood K5) was better correlated with adult K1 and K5 (0.28 vs 0.11 for K1; 0.33 vs 0.25 for K5 at age 11 to 13 years). K4 was also shown to have a higher odds ratio for predicting adult hypertension than K5 (1.57 vs 1.14 at age 11 to 13 years). CONCLUSIONS During childhood K4 is a more reliable measure of diastolic blood pressure than K5. K4 diastolic blood pressure measured in childhood is a better predictor of adult hypertension.