Endometrial carcinomas are histologically classified as endometrioid, assumed to originate from hyperplastic endometrium, or non-endometrioid carcinomas, assumed to originate from atrophic endometrium. However, both on a histological and a molecular level there are indications that there are more carcinoma types and carcinogenetic pathways. This study aims to analyze endometrial carcinogenesis on a molecular level. The presence of known KRAS, PIK3CA, AKT1, CTNNB1, BRAF, EGFR and NRAS mutations was studied in proliferative, atrophic and hyperplastic endometrium, endometrioid and serous carcinomas, and the endometrium next to these carcinomas, using single molecule Molecular Inversion Probes. Mutations were found in 9 (15%) of the 62 non atypical, and in 6 (18%) of the 34 atypical hyperplasia cases. In comparison, mutations were found in 1 (3%) of the simple, and 8 (30%) of the 27 complex hyperplasia cases. In 12/22 (55%) endometrioid carcinomas, a mutation was found. The KRAS gene was most often mutated in carcinomas next to hyperplastic endometrium, whereas PIK3CA and CTNNB1 mutations were found in endometrioid carcinomas with adjacent atrophic endometrium. Complex hyperplasia rather than atypical hyperplasia appears to be the most important lesion in the carcinogenesis of endometrioid carcinomas, and KRAS, PIK3CA and CTNNB1 mutations appear to play an important role in this process. Carcinogenesis of endometrioid carcinomas next to hyperplasia seems to be different to that of those next to atrophia. The value of these findings in managing endometrial hyperplasia and carcinoma should be studied.