The effect of deficit of the FMR1-gene product (FMRP) on physical phenotype was investigated using a robust modification of the maximum likelihood estimators for pedigree data. The approach is a powerful method of examining genotype-phenotype relationships because it adjusts for intra-familial variation, and the robust modification allows violation of distributional assumptions in the data to be overcome by objectively down-weighting unusual observations. The data on 19 age- or height-adjusted physical measures including head, trunk and limb measures and height and weight from 110 extended fragile X families (including 185 fragile X males and females and 120 normal relatives) were related to the FMRP levels assessed in peripheral blood lymphocytes. A significant interaction between FMRP and age was also included in the models for some measures. The results have demonstrated a linear effect of progressively reducing levels of FMRP on the values of a majority of physical measures considered in the study. The most evident effect of FMRP deficit in sexes combined was in decreasing body height and limb length, and in increasing head height and the degree of connective tissue involvement (measured by the middle finger extension angle). Heritability estimated from the complex FMRP models showed the highest values for height and limb length, and the lowest for weight, finger extension angle and some facial measures. On the basis of the present data, a possible mechanism by which the FMRP deficit impacts physical phenotype is discussed.