BACKGROUND. The objective of this study was to compare the accuracy of disease progression prediction of the molecular genetics and morphometry-based Endometrial Intraepithelial Neoplasia (EIN) and World Health Organization 1994 (WHO94) classification systems in patients with endometrial hyperplasias. METHODS. A multicenter, multivariate analysis was conducted on 477 patients with endometrial hyperplasia who were required to have a 1-year minimum disease-free interval from the time of the index biopsy (1-18 years of follow-up). The results from that analysis were compared with the results from 197 patients who had < 1 year of follow-up. RESULTS. Twenty-four of 477 hyperplasias (5.0%) progressed to malignant disease over an average of 4 years (maximum, 10 years). According to the WHO94 classification, 16 of 123 atypical hyperplasias (13%) and 8 of 354 nonatypical hyperplasias (2.3%) progressed (hazard ratio [HR] = 7). Twenty-two of 118 EINs (19%) and 2 of 359 non-EINs (0.6%) progressed (HR = 45). EIN was prognostic within each WHO94 subcategory. Progression rates were 3% in simple hyperplasias, 22% in complex hyperplasias, 17% in simple atypical hyperplasias, and 38% in complex atypical hyperplasias with EIN, compared with progression rates of 0.0-2.0% in all hyperplasias if EIN was absent. EIN detected precancerous lesions (sensitivity, 92%) better than WHO94 atypical hyperplasias collectively (67%) or complex atypical hyperplasias alone (46%). In a Cox regression analysis, EIN was the strongest prognostic index of future endometrial carcinoma. The same was true for patients with < 1 year of follow-up (HR for EIN, atypical hyperplasia, and complex atypical hyperplasia: 58, 7, and 8, respectively). CONCLUSIONS. The EIN classification system predicted disease progression more accurately than the WHO94 classification and identified many women with benign changes that would have been regarded as high risk according to the WHO94 classification system.