We have identified two females who are mosaic for an ATRX mutation. One case, in whom the mutation was undetectable in peripheral blood and buccal cells, has two affected sons and is therefore presumed to be a germline mosaic. In another case, the ATRX mutation is weakly detectable in the peripheral blood but only one of her three children who share the diseaseassociated haplotype carries the mutation and therefore it is concluded that she is a gonosomal mosaic. These cases provide the first molecular evidence for the occurrence of post-zygotic mutation in X-linked α thalassaemia mental retardation syndrome. The possibility of germline mosaicism must therefore be considered in the genetic counselling of ATR-X families.