From 2,874 school children participating in the 1971 and 1973 Muscatine Coronary Risk Factor Survey, we selected three groups of index cases for detailed family study: the HIGH group (n-56), with cholesterol levels greater than the 95th percentile twice; the MIDDLE group (n=46), cholesterol levels between the 5th and 95th percentile; and the LOW group (n=46), cholesterol levels less than the 5th percentile twice. Coronary mortality determined from death certificates was increased in the young relatives (ages 30-59) of the HIGH group index cases, as follows: twofold excess in HIGH male relatives compared with the MIDDLE or LOW group (p < 0.05); tenfold excess in the HIGH female relatives compared with the MIDDLE and LOW group combined (p < 0.01). After correction for years at risk, there was an approximately twofold significantly-increased coronary mortality. Stroke mortality was higher, although not significantly, in the older relatives (ages >60) of the HIGH index cases. Cancer mortality was not significantly different among the relatives of the three groups of index cases. This study indicates that school children's cholesterol levels cluster with those of their family members and that persistent hypercholesterolemia in children identifies families at risk for coronary artery disease.