Although population-based studies have unequivocally reported an increased risk of ovarian cancer in women with endometriosis, the biological evidence supporting the idea of endometriosis as a preneoplastic condition is scanty and not well substantiated. The fundamental features of human neoplasms (monoclonal growth, genetic changes, mutations in tumour suppressor genes and replicative advantage) have been evaluated in endometriotic lesions but results obtained are discordant. It is plausible that ectopic glands may expand monoclonally but the entity of this phenomenon is debated. According to some allelotyping studies, from one-third to one-half of endometriosis lesions would harbour somatic genetic changes in chromosomal regions supposed to contain genes involved in ovarian tumourigenesis, especially for the endometrioid histotype. These findings would be consistent with the progression model for carcinogenesis from the benign precursor to ovarian cancer but they could not be unequivocally replicated. Gene mutational studies are rare in this context. A single group has found missense mutations and deletions of PTEN gene in about 20% of ovarian endometriotic cysts. Moreover, in a model of genetically engineered mice harbouring an oncogenic allele of K-ras resulting in benign lesions reminiscent of endometriosis, a conditional deletion of PTEN caused the progression towards the endometrioid tumour. Based on these data, the causal link between endometriosis and ovarian endometrioid/clear cell carcinomas remains to be defined both in terms of entity of association and of underlying molecular mechanisms.