We studied lipids and lipoprotein cholesterols in 39 children (26 boys, 13 girls) with severe migraine, to examine the hypothesis that primary and familial lipoprotein abnormalities might be associated with or predispose children to the migraine syndrome. Each of the children, 4 to 20 years of age, had severe migraine, leading to pediatric neurologic referral and therapy. Twenty-five of the 39 probands (64%) had a first degree relative with severe migraine, and 18% had a second degree relative with severe migraine. In 11 of the 39 kindreds (28%), there was a family history of premature myocardial infarction and/or cerebral vascular accident (<age 55 years), involving one grandparent from each of ten kindreds and one parent in the 11th kindred. In nine of the 26 boys, low-density lipoprotein cholesterol (LDL-C) levels were greater than or equal to the age-, sex-, race-specific 90th percentile, and in three of these nine children, there was at least one additional first degree relative also having a primary top decile LDL-C level, consistent with the presumptive diagnosis of familial hypercholesterolemia. The finding of more than three times as many boys with migraine headache having top decile LDL-C than expected (9 v 2.6) was significant (x2 = 17.5, P <.01). Also, there were six boys having bottom decile levels of high-density lipoprotein cholesterol (HDL-C); all six came from kindreds with at least one first degree relative also having bottom decile HDL-C. The finding of more than two times as many boys with migraine having bottom decile HDL-C than expected (6 v 2.6) was significant (x2 = 4.94, P < .05). Of the 13 female pediatric probands, two had top decile LDL-C and two had bottom decile HDL-C and came from families with at least one additional first degree family relative also having a primary and similar dyslipoproteinemia. Our observations suggest that the clinical diagnosis of severe migraine in childhood should lead to measurement of lipids and lipoprotein cholesterols, particularly in boys, because they represent a cohort with a disproportionate number of hyper-β- and hypo-α-lipoproteinemic subjects. We speculate that primary and familial lipoprotein abnormalities, particularly those involving high levels of LDL-C and/or low levels of HDL-C, may be etiologically related to migraine, perhaps related to platelet hyperaggregability, and/or increased likelihood of cerebral vascular instability.