Background Early stage (FIGO stage Ⅰ-Ⅱ) endometrioid endometrial adenocarcinoma (EEA) is very common in clinical practice.However,patients with the early stage EEA show various clinical behaviors due to biological heterogeneity.Hence,we aimed to discover distinct classes of tumors based on gene expression profiling,and analyze whether the molecular classification correlated with the histopathological stages or other clinical parameters.Methods Hierarchical clustering was performed for class discovery in 28 eady stage EEA samples using a special cDNA microarray chip containing 492 genes designed for endometrial cancer.Correlations between clinicopathologic parameters and our classification were analyzed.And the significance analysis of microarrays (SAM) array was used to identify the signature genes according to the tumor grade and myometrial invasion.Results Three tumor subtypes (subtypes Ⅰ,Ⅱ and Ⅲ) were identified by hierarchical clustering,each subtype had different clinicopathological factors,such as tumor grade,myometrial invasion status,and FIGO stage.Moreover,SAM analysis showed 34 up-regulated genes in high grade tumors,and 38 up-regulated genes and 1 down-regulated in deep myometrial invasive tumors.The overlap genes between these two high-risk factors were markedly up-regulated in subtype Ⅰ,but down-regulated in subtype Ⅲ.Conclusion We have identified novel molecular subtypes in early stage EEA.Differential gene signatures characterize each tumor subtype,which could be used for recognizing the tumor risk and providing a basis for further treatment stratification.