During the late 1970s, there was a dramatic reduction in postmenopausal estrogen use in the United States, which may have reflected concern over a well-publicized postmenopausal estrogen-endometrial cancer link. The authors studied 310 postmenopausal women in a defined population over the period 1974-1981 to evaluate whether hysterectomy and certain other characteristics predicted change in postmenopausal estrogen use status during this period and, as a secondary issue, whether women who subsequently began postmenopausal estrogen use had different characteristics prior to use, an important question in the evaluation of the relation of postmenopausal estrogen use to morbidity and mortality from cancer, cardiovascular disease, or other diseases in observational studies. The only strong predictor of whether postmenopausal estrogen use would be discontinued was the presence of an intact uterus. Women who discontinued postmenopausal estrogen use were also somewhat older and heavier than those who continued, but were otherwise quite similar on a wide range of variables, including risk factors for and the presence of various chronic diseases. Similarly, the absence of a uterus was the only strong predictor of the initiation of postmenopausal estrogen use. Thus, concern about a possible postmenopausal estrogen-endometrial cancer link appeared to have been the major determinant of change in postmenopausal estrogen use in this time period. In the secondary analysis, variables other than hysterectomy did not discriminate between women who initiated postmenopausal estrogen use versus those who did not report use of postmenopausal estrogens, suggesting that a broad range of other characteristics was not a priori different in these two groups.