A sex-specific familial correlation model was used to assess the herit
able contributions to several measures of body composition in 86 seden
tary white families participating in the HERITAGE Family Study. For th
is study, sedentary families mere recruited, tested for a battery of m
easures, endurance exercise trained for 20 weeks, and remeasured. This
sample is unique in that activity level was controlled for in these f
amilies at baseline measurement. In this report, three body compositio
n variables measured at baseline were analyzed, two indexing adiposity
(total subcutaneous fat based on eight skinfold measurements [SF8] an
d percent body fat measured by underwater weighing techniques [%BF]) a
nd one assessing fat free mass ([FFM] derived from underwater weighing
). The maximal heritabilities for SF8 (34%) and %BF (62%) were consist
ent with those reported in previous studies. There were no sex nor gen
eration differences in the familial correlations, and the spouse corre
lation was significant, consistent with the hypothesis that the famili
al aggregation reflects genetic and familial environmental factors. Ho
wever, the results for FFM were very different. The most parsimonious
pattern of familial resemblance was consistent with mitochondrial inhe
ritance (i.e., mother-offspring and sibling correlations were equal an
d were larger than those for spouse and father-offspring pairs). Under
the mitochondrial hypothesis, 39% of the variance was accounted for b
y familial/genetic effects. However, under a nonmitochondrial hypothes
is, which could not be ruled out, 65% of the FFM phenotypic variance w
as accounted for by familial/genetic factors. This high heritability l
evel, as compared with results from previous studies, is consistent wi
th the hypothesis that activity may constitute an important environmen
tal determinant of FFM. These alternative hypotheses for FFM warrant f
urther investigation using complex multilocus-multitrait segregation m
odels, which allow for major genetic, polygenic, and environmental sou
rces of variance, as well as interactions among them.