Insulin-like growth factor II mRNA-binding protein 3 (IMP3) is an oncofetal protein highly expressed in fetal tissue and malignant tumors but rarely found in adult benign tissues. The aim of this study is to determine the expression of IMP3 in benign endometrium, endometrial cancer, and its precursor lesions, trying to see whether IMP3 has any diagnostic usage. Two hundred ninety-eight endometrial samples were examined for IMP3 expression by immunohistochemistry. These included benign endometrium (n=68), atypical hyperplasia or endometrial intraepithelial neoplasia (n=35), endometrial glandular dysplasia (n=21), endometrial intraepithelial carcinoma (n=18), endometrioid carcinoma (n=70), mucinous carcinoma (n=8), serous carcinoma (n=51), clear cell carcinoma (n=12), and other malignancies (n=15). Maturational patterns in the 68 benign endometrial samples included atrophic (n=12), proliferative (n=18), secretory (n=14), menstrual (n=8), and gestational (n=16). Most of the carcinomas were histologically pure; where mixed, the second component constituted <10% of the total tumor volume. The extent and intensity of IMP3 expression was semiquantitatively determined and scored for all samples. A renal cell carcinoma with known IMP3 expression was used as positive control for each immunohistochemistry run. Among the malignant cases, IMP3 expression was predominantly found in endometrial serous carcinoma and its putative precursor lesions, with 3 (14%) of 21 endometrial glandular dysplasia, 16 (89%) of 18 serous endometrial intraepithelial carcinoma, and 48 (94%) of 51 serous carcinomas (P<0.001). In contrast, the frequency of IMP3 expression was significantly lesser in nonserous malignancies with 0 (0%) of 35, 5 (7%) of 70, 0 (0%) of 8, 3 (25%) of 12, and 5 (33%) of 15 positive expression rates in atypical hyperplasia or endometrial intraepithelial neoplasia, endometrioid, mucinous, clear cell carcinomas, and other malignancies, respectively. The IMP3 staining was universally cytoplasmic, with diffuse staining of strong intensity in serous carcinomas, whereas staining was typically patchy and of moderate or weak intensity in nonserous malignancies. Among the benign endometrial samples, decidualized endometrial stroma showed 100% positivity for IMP3. The remaining samples were negative, with the exception of a few weakly proliferative glands in 3 (5%) of 68 cases that showed focal weak immunoreactivity of IMP3. The trophoblasts in the first trimester chorionic villi were also diffusely positive, which was consistent with previously reported findings. We conclude that expression of IMP3, a newly identified cytoplasmic marker, is closely associated with type II endometrial cancer. It seems that IMP3 expression is associated with an aggressive histologic phenotype among endometrial neoplastic lesions. Strong and diffuse IMP3 expression is highly sensitive for endometrial serous and clear cell carcinomas including their putative precursor lesions. Therefore, IMP3 may be a useful diagnostic marker in the assessment of endometrial cancers and their precursor lesions, particularly when the amount of available tissue material is limited and a concern of type II cancer arises. High frequency of IMP3 expression is present in decidualized endometrial stroma of gestational endometrium and chorionic villi in early pregnancy. Although the significance of the latter finding remains unclear, the differential diagnosis between decidual changes and endometrial serous carcinoma is rarely problematic.