We hypothesize that endometrial carcinoma and their precursors share a monoclonal growth pattern and tested this thesis with archival paraffin-embedded tissues using a polymerase chain reaction-based assay for non-random X chromosome inactivation. Of the 10 well-differentiated endometrial adenocarcinoma cases with heterozygous markers (HUMARA, X-linked androgen receptor gene), 9 had skewed X inactivation consistent with a monoclonal process, and one contained a structurally altered HUMARA gene. X inactivation skewing similar to that of the tumor was seen in matched control polyclonal tissues of 4 (of 9) cases, caused by the small number of endometrial stem cells at the time of embryonic X inactivation. When the polymerase chain reaction assay was applied to four potential endometrial precancers (atypical endometrial hyperplasia) and matched control tissues, two were inconclusive, and two were found to be monoclonal. We conclude that 1) it is essential to include polyclonal control tissues in X inactivation analyses to determine whether skewing is a specific indicator of monoclonality; and 2) endometrial adenocarcinomas and some putative precancers, atypical endometrial hyperplasia, are monoclonal.