Two problems were the focus of this study. (1) Is precancer and/or invasive cancer of the human cervix a poly- or monoclonal proliferation of neoplastic cells? (2) Are simultaneously present precancers and cancers of the cervix clonally related, or do they arise independently? Microdissection of 37 neoplastic lesions with different degrees of histologic severity in 22 patients followed by polymerase chain reaction-based analysis of X-chromosome inactivation was used as a principal method. Invasive cancers were interpreted as monoclonal because samples invariably showed monoclonal signals. In two thirds of these cases, simultaneously present precursors had the identical X-chromosome inactivation pattern, but in one third the pattern was different. Polyclonality was seen in a subgroup of precursors, where there was no simultaneous presence of invasive cancer. In contrast, when invasive cancer was present, no precursor signaled polyclonality. Data taken together indicate that the pathogenesis of cervical cancer is probably even more complicated than that of other cancers involving selection of subclones from originally polyclonal precursors and possibility of coexistence of precursors of different monoclonal composition. The study also observed that a large field of normal cervical squamous epithelium (approximately 500 basal squamous epithelial cells) with nonrandom X-chromosome inactivation was present. It remains to be further investigated whether this phenomenon represents an embryologic lyonization pattern of X-chromosome inactivation or postembryologic clonal expansion of submorphologically transformed cells.