Longitudinal data from twin pairs may be used to determine how the genetic
effects influencing a quantitative trait change with age. Here a model for
mixed longitudinal data of Huggins and Loesch [1998] on unrelated individua
ls is extended to twin studies. The model is fitted using robust statistica
l methods and a bootstrap procedure is proposed to estimate the percentiles
. The method is applied to longitudinal twin data on body mass index in mal
e and female twin pairs aged 5-18 years. (C) 2000 Wiley-Liss, Inc.