If the mother is the fragile X gene carrier, her daughters (and sons) with the mutation are at high risk of mental retardation. If the father is the (clinically unaffected) carrier, his daughters are normal. This is strong evidence for a maternal effect. The decreased penetrance and variable expressivity in fra(X) offspring of carriers could be related, at least in part, to variabile expression or availability of some maternal factor between pregnancies. We hypothesize a maternal effect in fra(X), with variability in intelligence of heterozygotes and hemizygotes mediated mainly by the maternal uterus or placenta by virtue of different patterns of lyonization in those tissues between pregnancies. If the mother is a carrier, the maternal placenta could develop with a skewed proportion of the normal or the fra(X) genetically active. Each female or male embryo could be exposed to very different environments with respect to genetic activity of the fra(X) chromosome, depending on the site of implantation within the uterus. If the father contributes the fra(X), the intrauterine environment is invariably normal and so are the daughters. Modifiers of the intrauterine effect could include lyonization patterns in tissues of the carrier fetus, and preferential inactivation of the paternal X in extra-embryonic tissues. The ultimate phenotype of the developing heterozygote and hemizygote may be determined by a threshold effect and interaction between the maternal genotype, the placental genotype, and the fetal genotype. The possibility of maternal effect is testable and has implications for treatment.