We set out to determine the relative timing of loss of DNA mismatch repair and KRAS2 mutation in endometrial tumorigenesis. We studied endometrial carcinoma (CA) and synchronous atypical endometrial hyperplasia (AEH), the premalignant precursor of endometrial cancer. Carcinoma and hyperplasia were investigated for loss of mismatch repair as evidenced by microsatellite instability (MSI) and for KRAS2 mutations. Endometrial cancers previously shown to be MSI-positive were evaluated for KRAS2 codon 12 and 13 mutations. DNA was isolated from foci of AEH concomitant with, but physically remote from, the cancers by use of tissues prepared by laser capture microdissection (LCM). The AEH DNAs were then assessed for MSI and KRAS2 mutations. Of 210 endometrial CAs investigated, 51 (26%) were MSI-positive, and among those, 21 (41%) arose concomitantly with AEH. Of 41 foci of AEH (mean, two foci per patient) investigated, 34 (83%) were MSI-positive. KRAS2 mutations were seen in 5/51 (10%) MSI-positive carcinomas. From the five patients informative for both KRAS2 mutation and MSI, 10 foci of AEH were available for investigation. All 10 AEH specimens (100%) were MSI-positive, and six (60%) had the KRAS2 mutation present in the coexisting CA. The observation that some MSI-positive AEH specimens lack the KRAS2 mutation seen in the coexisting CA supports a model in which loss of DNA mismatch repair precedes KRAS2 mutation. However, in addition to the absence of KRAS2 mutations in AEH, we discovered mutations in LCM hyperplasia and carcinoma specimens that were not present in the portion of the cancers originally investigated. These discordant genotypes suggest genetic heterogeneity in endometrial hyperplasia and concomitant cancer.