Endometrial adenocarcinoma (EA) with myometrial involvement limited to foci of adenomyosis has been associated with a better 5-year survival than EA with myometrial invasion at the corresponding depth. We identified 23 cases of stage I EA diagnosed between 1975 and 1981 in which myometrial involvement was confined entirely to adenomyotic foci. Histopathological criteria used to determine adenomyotic involvement by EA included presence of endometrial stroma; presence of adjacent benign "marker" glands to indicate partial replacement of adenomyosis; either bulging expansion of the endomyometrial junction by EA or a smooth rounded contour of entirely intramyometrial tumor nests; and absence of peritumoral desmoplasia or stromal loosening around such foci. In any one case no single criterion was sufficient to differentiate adenomyotic involvement from true invasion; however, none of the cases showed the last phenomenon. Adenomyotic involvement extended to the inner third of the myometrium in 15 cases, the middle third in 6 cases, and the outer third in 2 cases. Twenty-one cases were pure adenocarcinoma, with one adenocarcinoma with squamous differentiation (adenoacanthoma) and one adenosquamous carcinoma; 18 cases were FIGO grade 1 and 5 were FIGO grade 2. Adenomyosis containing atypical hyperplasia was seen in 13 cases, suggesting that EA may arise de novo in adenomyosis at variable levels in the myometrium. Current follow-up data were available for all patients, with 19 presently alive and free of disease. Four died of unrelated causes, three of whom had inner third involvement and one, middle third involvement. Twelve patients were treated with preoperative or postoperative radiation. This study supports previous smaller series suggesting that cases of EA in which myometrial involvement is limited to adenomyosis have a better prognosis than those with true myometrial invasion at an equivalent level and that adenocarcinoma may arise de novo in adenomyosis. © 1990.