235 cases of primary endometrial adenocarcinoma (AC) (age range, 37-94; mean age, 61 years) were collected during the period 1980-1992. Hysteroscopic examination of both the endometrial cavity and the cervical canal was performed in every patient prior to hysterectomy, and evaluations of cancer extension in the endometrium (focal: 97 pts; partial: 82 pts; massive: 47 pts; unevaluable: 9 pts) and of endocervical involvement (positive: 45 pts) were compared to the histological findings and survival rates. The chi 2 test was used for statistical analysis, and statistical significance was considered where the p value was < 0.05. Endometrial extension was poorly related to the depth of myometrial invasion (M1 = depth of invasion to < 1/2 myometrium, M2 = invasion to > 1/2 myometrium): focal AC: M1 57.8%, M2 42.2%; partial AC: M1 40.2%, M2 59.8%; massive AC: M1 51.1%, M2 48.9%; (p = 0.5). Endocervical involvement was unrelated to endometrial extension. No correlation was found between AC histological grade (G1-G3) and entity of endometrial extension, whereas grade showed a significant correlation with myometrial invasion (G1 M1: 69.1%; G3 M1: 41.0%; p = 0.002) and survival rates (G1 90.4%, G2 88.5%, G3 69.4%; p = 0.01). Five-year survival figures showed no evident correlation with cancer extension (focal AC: 86.5%; partial AC: 87.8%; massive AC: 86.3%; p = 0.9) whereas myometrial invasion showed a statistical significance (M1: 91.4%, M2: 79.7%; p = 0.03). Three patterns of invasion were defined: pushing (P), infiltrative (I) and diffuse (D) isolated cells. There were significant differences between the various growth patterns and survival rates (P 90.7%; I 84.3%; D 45.4%; p = 0.0001). False negative rate of the hysteroscopic diagnosis of cervical involvement was 7.9% (18 cases); however, in 6 of these cases only deep cervical invasion was found.