To determine the relationship between premature ovarian failure (POF) and skewed X-chromosome inactivation (XCI), karyotype, and XCI status in 43 patients with POF (group I) and 43 age-matched control women with regular menstrual cycles (group II) were evaluated. Evaluation of XCI status was based on the CAG triplet repeat polymorphism assay in the androgen receptor gene after sodium bisulfite treatment of DNA samples, and XCI patterns were classified as random (XCI < 70% skewing) or skewed (> or =70%). Furthermore, skewed XCI was classified under three different thresholds (> or =70, > or =80, or > or =90%). Karyotyping by G-banding and fluorescence in situ hybridization (FISH) on peripheral blood lymphocytes showed that one patient in group I had a deletion of Xq22, and another was 47,XXX. The frequency of low-level 45,X/46,XX mosaicism was nearly equal in both groups. In women without any X-chromosomal aberrations, the incidence of skewed XCI in group I was significantly higher than in group II on all threshold levels. Furthermore, extremely skewed XCI (> or =90%) was observed only in group I. These results indicate that POF may be caused by some underlying genetic disorders, which may induce skewed XCI.