The hypothesis of X-linked new mutations which might cause early abortions of hemizygous male fetuses and a dominant phenotype in heterozygous females seems the most likely genetic explanation of the Rett syndrome. This hypothesis can be reconciled with the normal sex ratio observed in sibships of patients and with the rare recurrence of this disorder in sibs or half-sibs. The latter observation can be explained by germinal mosaicism in one of the two parents. Since in 14 patients no association was found with any particular fragile site or chromosome rearrangement, we propose to map the mutated gene (or loci) on the X through a strategy based on the reconstruction of X-linked haplotypes consisting of DNA polymorphisms, and on the identification of possible crossovers in affected sisters.