The patterns of X chromosome inactivation and mutations of PTEN and K-ras were evaluated in cases of endometrial hyperplasia to determine the presence of potentially premalignant neoplastic versus polyclonal benign cell populations. Endometrial glandular epithelial cells were collected by laser capture microdissection, and genomic DNAs were extracted. Following treatment with the methylation sensitive restriction endonuclease Hha I, polymerase chain reaction amplification was performed targeting a highly polymorphic short tandem repeat of the human androgen receptor gene (HUMARA). PTEN and K-ras gene mutations were evaluated by analysis of single-strand conformation polymorphism. Two pathologists performed histologic diagnosis of the lesions independently. Monoclonal composition was demonstrated in 13 of 15 (87%) endometrial hyperplasias with atypia and 17 of 31 (55%) complex hyperplasias without atypia. Cytological atypia is significantly associated with the clonal status of the endometrial hyperplasia (13/15 vs 17/31, P = 0.049). In contrast, all 14 normal endometrial tissue samples were polyclonal. PTEN gene mutations were detected in 4 of 13 (30%) monoclonal endometrial hyperplasias with atypia and 2 of 17 (12%) monoclonal endometrial hyperplasias without atypia but were not detected in polyclonal endometrial hyperplasias, with or without atypia. K-ras gene mutations were present in 3 of 13 (23%) monoclonal endometrial hyperplasias with atypia but not in 2 cases of polyclonal endometrial hyperplasia with atypia or in 26 cases of endometrial hyperplasia without atypia. K-ras mutation is thus significantly more frequently found in endometrial hyperplasias with atypia than those without atypia (3/15 vs 0/31, P = 0.030). This study indicates that most cases of endometrial hyperplasia with atypia and a high proportion of cases of endometrial hyperplasias without atypia originate from a single progenitor cell, possibly as a result of genetic alterations, rather than as a result of benign reactive processes.