The relationship of hyperplasia to carcinoma of the endometrium has therefore an etiologic and a clinical aspect. 1.I. Etiologic relationship: The position of endometrial hyperplasia as a precancerous lesion rests upon the following evidence: 1.1.1. Morphologic similarity: The weight given such evidence is largely dependent upon individual conception of form and is not susceptible to direct proof. A review of 85 cases of endometrial hyperplasia suggests, however, that this entity consists in reality of a series of types varying from those which differ little from normal endometrium to others which closely resemble certain differentiated carcinomas.1.2.2. Biologic similarity. The frequent association of endometrial hyperplasia with adenomyosis or the invasion of the muscularis by mucosal tissue and the tendency of the disease to return after curettage are perhaps to be interpreted as representations in miniature of two of the chief properties of malignancy, infiltration and recurrence.1.3.3. Transformation of hyperplasia into carcinoma, as indicated by a change in the character of the tissue obtained in successive curettings in the same patient, has been reported in the literature in at least six instances. Such cases must, however, be regarded critically because of the frequent lack of precision in the use of the term hyperplasia and the possibility that undetected carcinoma may have been present in the uterus at the time of the first operation. The only two instances occurring among 85 cases treated for hyperplasia of the endometrium at the Roosevelt Hospital from 1925 to 1929 of apparent transformation to carcinoma were dependent upon such errors.A review of 122 histories in cases of corpus carcinoma brings out the fact, however, that very many of these women had at some time before their final operation for cancer been under treatment for abnormal uterine bleeding. These cases may be divided as follows: 1.3.1.a. Cases in which endometrial hyperplasia almost certainly preceded carcinoma (Cases 5, 9).1.3.2.b. Cases in which the time and characteristics of the previous bleeding make it probable that a period of benign endometrial disease preceded the development of the malignant tumor (Cases 10, 11, 18, 20).1.3.3.c. Cases in which the earlier condition was obviously regarded by the clinician or pathologist as benign but which subsequent events indicate was very likely malignant (Cases 3, 4, 6, 7, 8, 17, 19, 21).1.3.4.d. Cases in which the earlier bleeding was later definitely proved to be due to an unrecognized malignant growth (Cases 1, 2, 12).1.4.4. The association of diffuse endometrial hyperplasia and carcinoma in the same uterus has been reported by previous observers and was noted in five instances (Cases 24, 25, 26, 27, 28) in the present cancer series. Similar proliferative changes were observed in 4 cases (Cases 1, 29, 30, 31) in women over sixty, in which, on account of the patient's age the unqualified use of the term endometrial hyperplasia has been withheld. In two further instances the carcinoma occurred with hyperplastic glands which were probably a part of an adenomatous polyp. Finally there were 9 cases of carcinoma associated with areas of invasion of the superficial muscularis by benign glands, constituting a condition termed adenomyosis and indicating abnormal properties in the basal endometrial glands.Although in a total of 152 cases there were only 15 with definite histologic evidence of an associated hyperplastic condition of some type, one cannot speak in terms of percentages, since it is obvious that many cases are so advanced that possible benign preexistent lesions have been completely displaced by carcinoma.2.II. Clinical relationship. Several cases in this series indicate that even with thoughtful handling, endometrial cancer may be mistaken for a benign condition. These cases lead to the following more or less obvious conclusions: 2.1.1. Postmenopausal bleeding from the uterine canal even if limited to a single attack should always be treated by curettage. No period of observation is sufficient to give security because further clinical evidence of carcinoma may not appear for even ten years after such an attack (Cases 17, 18, 19).2.2.2. Curetted material, no matter how scant, may be carcinomatous, and it is never justifiable to dispense with microscopic examination (Case 7).2.3.3. An incomplete curettage is not satisfactory as a diagnostic measure, for a small carcinoma may be missed by the instrument.2.4.4. A single microscopic section of curettings is not sufficient to rule out cancer in suspicious cases because the microtome may not cut the particle containing the growth (Cases 1 and 26).2.5.5. The histologic differentiation of hyperplasia from certain types of carcinoma requires at times considerable experience and the examination of multiple sections. The possibility of errors in diagnosis is very real and mistakes may lead to disastrous results (Cases 2, 12, and possibly 3, 4, and 6).Whether from a practical standpoint hyperplasia is to be regarded as precancerous and treated as such must remain an open question. The relative frequency of hyperplasia undoubtedly indicates that the individual patient with the disease is reasonably safe. Nevertheless it appears that when the hyperplasia is at all marked, the possibility of a predisposition to the development of cancer should be considered and the case regarded with the same degree of suspicion now bestowed upon the diffuse forms of hyperplasia of the breast epithelium. In patients of the menopause age and older an adequate dose of radium is particularly indicated, certainly as the most efficient method of controlling bleeding, possibly as a prophylactic measure against the development of cancer.