A basic tenet of the Lyon hypothesis is that X inactivation occurs randomly with respect to parental origin of the X chromosome. Yet, nonrandom patterns of X inactivation are common - often ascertained in women who manifest recessive X-linked disorders despite being heterozygous for the mutation. Usually, the cause of skewing is cell selection disfavouring one of the cell lineages created by random X inactivation. We have identified a three generation kindred, with three females who have haemophilia A because of extreme skewing of X inactivation. Although they have both normal and mutant factor VIII (FVIII) alleles, only the mutant one is transcribed; and, they share an XIST allele that is never transcribed. The skewing in this case seems to result from an abnormality in the initial choice process, which prevents the chromosome bearing the mutant FVIII allele from being an inactive X.