Early identification of common familial dyslipidemias may prevent premature atherosclerotic disease. This study estimated the diagnostic values of few early childhood repeatedly deviant lipid samples by the knowledge of the parent's dyslipidemia. The first 7 years of age data of 353 children with their parents were evaluated from atherosclerosis risk-factor intervention study controls. Parents' low high-density-lipoprotein-cholesterol concentration (hypo-HDL-C), and high total cholesterol concentration-HDL-C (hyper-non-HDL-C) were defined. True hypo-HDL-C and hyper-non-HDL-C children were defined when their respective individual longitudinal means were beyond the appropriate lipid quintiles. Sensitivities, specificities, positive and negative predictive values of the early lipid samples were estimated with individual standard deviation models and bootstrap confidence. Hypo-HDL-C children proportions were 15.3% of all, and 20.9% of the children from the hypo-HDL-C parents (p = 0.26). Hyper-non-HDL-C children were 16.7% ofall and 31.8% of the children from the hyper-non-HDL-C parents (p=0.008). One early non-HDL-C sample in the highest quintile predicted 56% of the hyper-non-HDL-C children from healthy parents, but 83% of the hyper-non-HDL-C children from the hyper-non-HDL-C parents. Mean of three samples improved the latter prediction to 91 %. This showed that if hypercholesterolemic parent's child expressed repeatedly hyper-non-HDL-C, it predicts true dyslipidemia of the child. (C) 2008 Elsevier Ireland Ltd. All rights reserved.