U.T.M.D. Anderson Hospital cases filed as endometrial sarcoma were reviewed. Malignant mixed Müllerian tumors, Müllerian adenosarcomas, and poorly differentiated sarcomas of uncertain origin were excluded, and a minimum of 36 months follow-up was required. On pathologic examination, the remaining cases fell into two distinct groups: endometrial stromal sarcoma (11 cases) and poorly differentiated endometrial sarcoma (seven cases). The former were characterized histologically by small to medium-sized uniform cells and distinctive arterial vasculature resembling the spiral arteries to the normal endometrium; they were thus considered to exhibit endometrial stromal differentiation. The latter showed topographic configurations indicative of endometrial origin but did not demonstrate endometrial stromal differentiation; they were composed of generally larger cells with nuclear hyperchromatism and pleomorphism and had no distinctive vascular pattern. The mitotic rate was usually low in endometrial stromal sarcoma and high in poorly differentiated endometrial sarcoma, although two examples of the former had more than 20 mitoses per 10 high-power fields. All but one of the patients with endometrial stromal sarcoma were less than 50 years of age, whereas all of those with poorly differentiated endometrial sarcoma were older than 50 years. One patient with endometrial stromal sarcoma died of tumor, two were living with tumor at latest follow-up, and eight were clinically tumor-free. By contrast, six of those with poorly differentiated endometrial sarcoma died of tumor, all within 34 months. The extent of tumor was the major prognostic factor in both groups; the three patients with endometrial stromal sarcoma who were dead or living with tumor all had extrauterine extension at the time of hysterectomy (only one of the other eight did so) and all six who died with poorly differentiated endometrial sarcoma had myometrial invasion (the one survivor had a tumor limited to the endometrium). Other pathologic factors, including mitotic rate, had no prognostic significance in either category; of particular interest was that both patients who had endometrial stromal sarcomas with high mitotic rates were tumor-free. It is concluded that endometrial stromal sarcoma is a distinct tumor entity whose diagnosis should not depend on mitotic rate, and it is suggested that poorly differentiated endometrial sarcoma is closely related to malignant mixed Müllerian tumor, in view of clinical and pathologic similarities.