Children with familial hypercholesterolemia (FH) exhibit substantial variance of LDL cholesterol. In previous studies, family members of children with FH were included, which may have influenced results. To avoid such bias, we studied phenotype in 450 unrelated children with FH and in 154 affected sib-pairs. In known families with classical FH, diagnosis was based on plasma LDL cholesterol above the age- and gender-specific 95th percentile. Girls had 0.47 +/- 0.15 mmol/L higher LDL cholesterol, compared with boys (p = 0.002). Also in girls, HDL cholesterol increased by 0.07 +/- 0.03 mmol/L per 5 y (pfor trend = 0.005); this age effect was not observed in boys. The distribution of apolipoprotein (apo) E genotypes was not significantly different between probands, their paired affected siblings, or a Dutch control population. Carriers with or without one epsilon4 allele had similar LDL and HDL cholesterol levels. Within the affected sib-pairs, the epsilon4 allele explained 72.4% of the variance of HDL cholesterol levels (-0.15 mmol/L, 95% confidence interval -0.24 to -0.05, p = 0.003). The effect of apoE4 on HDL cholesterol differed with an analysis based on probands or on affected sib-pairs. The affected sib-pair model used adjustment for shared environment, type of LDL receptor gene mutation, and a proportion of additional genetic factors and may, therefore, be more accurate in estimating effects of risk factors on complex traits. We conclude that the epsilon4 allele was associated with lower HDL cholesterol levels in an affected sib-pair analysis, which strongly suggests that apoE4 influences HDL cholesterol levels in FH children. Moreover, the strong association suggests that apoE4 carries an additional disadvantage for FH children.