The origin of Rett syndrome has long been debated, but several observations have suggested an X-linked dominant inheritance pattern. We and others have pursued an exclusion-mapping strategy using DNA from a small number of familial Rett syndrome cases. This work resulted in the narrowing of the region likely to harbor the mutated gene to Xq27.3-Xqter. After systematic exclusion of several candidate genes, we discovered mutations in MECP2, the gene that encodes the transcriptional repressor, methyl-CpG-binding protein 2. Since then, nonsense, missense, or frameshift mutations have been found in at least 80% of girls affected with classic Rett syndrome. Sixty-four percent of mutations are recurrent C > T transitions at eight CpG dinucleotides mutation hotspots, while the C-terminal region of the gene is prone to recurrent multinucleotide deletions (11%). Most mutations are predicted to result in total or partial loss of function of MeCP2. There is no clear correlation between the type and position of the mutation and the phenotypic features of classic and variant Rett syndrome patients, and XCI appears to be a major determinant of phenotypic severity. Further research focuses on the pathogenic consequences of these mutations along the hypothesis of loss of transcriptional repression of a small number of genes that are essential for neuronal function in the maturing brain.