Gene therapy is unarguably the definitive way to treat, and possibly cure, genetic diseases. A straightforward concept in theory, in practice it has proven difficult to realize, even when directed to easily accessed somatic cell systems. Gene therapy for diseases in which the central nervous system (CNS) is the target organ presents even greater challenges and diverse vectors and brain delivery approaches are under investigation. We argue that in the case of the fragile X syndrome the approach most likely to have a chance of being effective should consist of a small, diffusible vector derived from the adeno-associated virus, carrying an FMR1 cDNA comprising the 5[prime] promoter region and the 3[prime] untranslated region of the gene, delivered to the entire brain by osmotic blood-brain barrier disruption. The approach can be tested in "Fmr1" knockout mice, although changes in their neurobehavioral abnormalities may be difficult to evaluate. A defect in the expression of GABA(A) receptors in these mice--if shown to be a direct consequence of the "Fmr1" defect--promises to be a more readily assessable marker of restored FMRp function on gene transfer.