Using the Princeton School District Family Study cohort, our specific aim was to estimate the prevalence of suspected familial ponderosity and leanness, to provide empirical risk estimates for the proportion of probands' first-degree relatives who were similarly affected, and to estimate the contributions of diseases, drugs and caloric intake to relative obesity and leanness. We studied 379 probands, 125 whites and 52 blacks from a random recall group, 147 whites and 55 blacks from a hyperlipidemic recall group. Suspected familial obesity and leanness were arbitrarily identified in those kindreds with at least two first-degree relatives in the same Quetelet index decile as the proband, top or bottom respectively. Suspected familial obesity was observed in 2.4 percent and 6 percent respectively of random and hyperlipidemic recall group whites. Suspected familial leanness was identified in 2.4 percent and 1.4 percent of random and hyperlipidemic recall whites and in 3.8 percent of randomly recalled blacks. Approximately twice as many as expected white first-degree relatives of top Quetelet index decile probands themselves had top decile Quetelet indices; approximately three times as many as expected first-degree relatives of bottom decile Quetelet index probands themselves had bottom decile Quetelet indices. Nineteen percent and 31 percent of top decile Quetelet index white probands from random and hyperlipidemic recall groups came from families where at least two other first-degree relatives were similarly obese; 18 percent and 20 percent of white random and hyperlipidemic recall group probands with bottom decile Quetelet indices had suspected familial leanness. Nearly all subjects with familial obesity or leanness had no overt metabolic or pharmacological explanations for their body habitus. Within-family clustering of hypertension was common in kindreds with suspected familial obesity and was absent in kindreds with suspected familial leanness. Marked within-family clustering of both obesity and leanness is useful diagnostically; therapeutic intervention to reduce obesity, to be most effective, should be family-wide in the many kindreds which share familial obesity.