Fragile-X syndrome is a major cause of mental retardation in humans. The X-inactivation imprinting model accounts for the unusual pattern of inheritance and expression of this syndrome. According to this model, the fragile-X mutation creates a local block to the attempted reactivation of the mutant X chromosome prior to oogenesis. This local block results in an "imprinted" fragile-X chromosome that is deleterious in males and in females for whom this chromosome is predominantly the active X chromosome. The imprinted state of the fragile-X mutation is inferred to be stable when transmitted by an imprinted female because the penetrance of the syndrome in sons of affected females is estimated to be 1.0. To provide a more precise estimate of the stability of the proposed fragile-X imprint, we have analyzed published pedigrees that include restriction fragment length polymorphism and cytogenetic data from sibships with mothers who are interpreted as having an imprinted fragile-X allele. We conclude that the fragile-X imprint was stable in 46 out of 48 female meioses. This analysis leads to a preliminary estimate of about 96% for the stability of the imprint through female meiosis. Two imprinted females had progeny who appeared to be carriers of a nonimprinted fragile-X allele. If this interpretation is correct, then reversion from the imprinted to the nonimprinted state, or "erasure," can occasionally occur when the mutant fragile-X allele is transmitted by an imprinted female. We discuss the genetic and epigenetic significance of possible female erasure. We request DNA and cytogenetic information from unpublished pedigrees to quantify further the stability, during female meiosis, of the proposed imprinted state of the mutant fragile-X allele.