Family resemblance for several measures of body fat and fat distribution was explored in the longitudinal Québec Family Study (QFS), including an overall measure of adiposity (body mass index, BMI), total subcutaneous fat (the sum of 6 skinfolds, SF6), and subcutaneous fat distribution (the trunk to extremity ratio, TER). Repeated measures were taken twice approximately 12 years apart. A longitudinal familial correlation model was used to assess familial resemblance at each of times 1, 2, and cross-time, and a univariate model was used for the change score. The change score was assumed to index the degree to which different familial factors impacted on the longitudinal resemblance, while the cross-time comparisons indexed similar familial factors across time. For BMI, the maximal heritability was 44 and 36% at times 1 and 2, respectively, 37% for the change score, and 33-43% for the cross-time comparison. While the etiology of the BMI familial effect at times 1, 2, and cross-time was assumed to be primarily polygenic, that for the change score was a function of cohort effects (environmental). For SF6, the maximal heritability (primarily genetic) was low at time 1 and for the change score (16%), but was nonsignificant at time 2 and cross-time. For TER, the maximal heritabilities were significant for each of times 1 (42%), 2 (40%), change score (59%), and cross-time comparisons (35-36%). In summary, simple univariate familial correlation analysis of the change scores and bivariate analysis of the longitudinal measures are useful in delineating the underlying factors leading to both change and stability across time.