As a result of X chromosome inactivation, females are mosaic for cell lineages in which either the paternal or the maternal X chromosome is active, and, if inactivation were random, each lineage should be present at approximately the same frequency. Detection of instances of non-random X inactivation can be important both clinically and for the study of X chromosome inactivation. Identification of a single-base polymorphism in an expressed region of the human XIST gene has permitted the development of a direct PCR-based assay for randomness of X inactivation. Oligonucleotide primers were designed, incorporating the variant base, and conditions established that allowed allele-specific PCR amplification. As the XIST gene is expressed only from the inactive X chromosome, differential amplification of the alleles in cDNA from heterozygotes can be used as an indicator of non-random inactivation. Using this assay, non-random X chromosome inactivation has been demonstrated in chromosomally abnormal cell lines and in lymphocytes from heterozygous, normal females. Virtually complete non-random X inactivation was also shown in a mother and her daughter, suggesting the existence of some familial factor affecting X chromosome inactivation.