The distinction of involvement of adenomyosis by endometrial carcinoma from endometrial carcinoma invading the myometrium can at times be difficult. This distinction, however, is important from the standpoint of staging, treatment, and prognosis because the outcome of carcinoma invading the myometrium as compared with involving adenomyosis is significantly worse. CD10 has been recently reported to be expressed by normal and neoplastic endometrial stromal cells. We therefore hypothesized that CD10 may be helpful in distinguishing carcinoma within adenomyosis from endometrial carcinoma directly invading the myometrium. Twenty-two cases of invasive endometrioid adenocarcinoma were identified from the surgical pathology files of the Johns Hopkins Hospital and consultation files of one of the authors (R.J.K.) and immunostained for CD10, desmin, and caldesmon. The pattern of staining was compared with five cases in which carcinoma was confined to adenomyosis. As a control, 14 cases of adenomyosis unassociated with carcinoma were included in the analysis. All 22 endometrial carcinomas that invaded the myometrium expressed CD10 to some extent in cells immediately surrounding the neoplastic glands. In 18, all of the invasive nests displayed CD10 in surrounding cells, but in four cases the staining was patchier, involving the surrounding cells of approximately 50-75% of the invasive nests. In four cases of myoinvasive carcinoma, the CD10-positive cells surrounding the nests of invasive carcinoma were also positive for desmin and caldesmon. In the remaining 18 cases with myoinvasive carcinoma, the cells surrounding the carcinomas failed to react with desmin and caldesmon. All five endometrial carcinomas involving adenomyosis displayed CD10 positivity in what appeared to be endometrial stromal cells surrounding the neoplastic glands. The stromal cells were negative for desmin and caldesmon. The control cases of adenomyosis were all positive for CD10, although in four cases the staining was patchy compared with 10 cases in which it was diffuse. Desmin and caldesmon were negative in all of these cases. Although CD10 identifies endometrial stromal cells in the endometrium and in adenomyosis and endometriosis, this study demonstrates that CD10 does not aid in distinguishing myometrial invasion of endometrial carcinoma from involvement of adenomyosis by endometrial carcinoma because the cells surrounding the tumor in the myoinvasive group express CD10.