Endometrial stromal sarcomas are low-grade malignant tumors that may pose a diagnostic challenge, especially when they are present in an extrauterine site. Owing to the presence of an arborizing vasculature and cells with an undifferentiated appearance, endometrial stromal sarcomas can be confused with several soft-tissue neoplasms. We studied 17 endometrial stromal sarcomas, eight hemangiopericytomas, 14 solitary fibrous tumors, and 16 synovial sarcomas immunohistochemically, detecting the following antigens: CD10, estrogen receptor, progesterone receptor, bcl-2, CD34, smooth muscle antigen, epithelial membrane antigen and cytokeratin (AE1/AE3). Most endometrial stromal sarcomas stained positively for CD10 (16/17), estrogen receptor (17/17), progesterone receptor (15/17), and bcl-2 (17/17). Staining with antismooth muscle antigen was seen in 11 of 17 cases of endometrial stromal sarcoma, with more intense staining seen in areas showing smooth muscle differentiation. Staining with AE1/3 was seen in four of 17 endometrial stromal sarcomas, with two of the positive cases containing epithelioid cells. None of the endometrial stromal sarcomas expressed epithelial membrane antigen or CD34. More than half of the hemangiopericytomas (4/8) and solitary fibrous tumors (9/14) cases demonstrated CD10 expression either focally or in a patchy cytoplasmic and membranous pattern. Hemangiopericytomas, solitary fibrous tumors, and synovial sarcomas did not express estrogen receptor. Four of eight hemangiopericytomas and seven of 14 solitary fibrous tumors also showed patchy progesterone receptor expression. CD34 expression was identified in six of eight hemangiopericytomas and 13 of 14 solitary fibrous tumors, but we did not find expression of CD34 in synovial sarcoma. Differences between endometrial stromal sarcoma and other soft-tissue tumors were detected for all of the immunohistochemical markers (P<0.05), except anti-bcl-2 and AE1/3. Antibodies against CD10 mark a substantial number of hemangiopericytomas and solitary fibrous tumors (albeit not diffusely) and should always be combined with antiestrogen receptor and CD34 when the differential diagnosis includes endometrial stromal sarcoma. Unlike estrogen receptor antibodies, progesterone receptor antibodies show at least focal nuclear staining in most hemangiopericytomas, solitary fibrous tumors and rare synovial sarcomas, and are not useful for this differential diagnosis. All endometrial stromal sarcomas expressed bcl-2, mostly in a diffuse pattern, but this did not distinguish between endometrial stromal sarcoma and mimics. We therefore recommend the use of a small antibody panel comprising anti-CD10, anti-estrogen receptor, and anti-CD34 to distinguish endometrial stromal sarcomas from tumors with a predominant hemangiopericytomatous growth pattern.