Rice T (Division of Biostatistics, Washington University School of Medicine, St Louis, MO, USA), Vogler G P, Perry T S, Laskarzewski P M, Province M A and Rao D L. Heterogeneity in the familial aggregation of fasting plasma glucose in five North American populations: The Lipid Research Clinics Family Study. International Journal of Epidemiology 1990, 19: 280-286.Heterogeneity in the familial aggregation of plasma glucose in five samples of the Lipid Research Clinics Family Study (LRC) was investigated using path analysis. This study was deemed appropriate since recent investigations reported a wide range of estimates for genetic and cultural factors. The path model incorporated a measured index of the familial environment in order to separate the effects of genes and environments in the nuclear family design, genetic and environmental heritabilities, spouse resemblance, sibling environmental effects, and parental cultural transmission. The methodology was completely general in allowing sample-specific, as well as pooled-sample, estimation of all or any subset of the model parameters. Genetic heritability estimates were heterogeneous, ranging from zero to 33% across the clinics. Environmental heritability (7%), spouse resemblance, non-transmitted sibling environmental effects, and parental cultural transmission were homogeneous across samples. No support was found for specific maternal effects, nor for intergenerational differences in cultural or genetic heritability. We conclude that the genetic and environmental heritabilities for plasma glucose in the LRC are consistent with the diverse reports by earlier investigators. In addition, we were able to exclude methodological differences as a cause of this heterogeneity. Furthermore, formal hypothesis tests suggest that the aetiology of this heterogeneity is genetic (and not cultural), taking the form of two distinct homogeneous patterns (one for no genetic effect, and one for a moderate genetic effect). Only formal heterogeneity tests of the type described here can detect these effects, and allow pooling of separate studies in order to obtain more precise estimates of the parameters of interest.