Prospective multicenter evaluation of the WHO classification and the morphometric D-score to predict endometrial hyperplasia cancer progression. In 132 endometrial hyperplasias WHO classification was performed by two experienced gynecologic pathologists. The D-score was assessed blindly by technicians in a routine diagnostic setting. Development of endometrial carcinoma during a 1-10-year follow-up was used as the end point. Eleven of 132 patients (8%), 10 of 61 (16%) atypical hyperplasias, and 1 of 71 (1%) nonatypical hyperplasias developed cancer. Twenty-six curettings had a D-score < or = 0 ("unfavorable" or endometrial intraepithelial neoplasia) of which 10 (38%) developed cancer. None of the 86 cases with a D-score > 1 ("favorable") and one of the 20 (5%) cases with 0 < D-score < or = 1 ("uncertain") developed cancer. Sensitivity of the D-score was 100%, specificity 82%, the positive and negative predictive values were 38% and 100%, respectively. These values are similar to those in three prior retrospective D-score studies but higher than the WHO values (which are 91%, 58%, 16%, and 99%, respectively). The D-score in endometrial hyperplasias is a more sensitive and specific marker for cancer prediction than the WHO classification, can be assessed in a routine clinical setting on standard hematoxylin and eosin sections (15-30 minutes per case), and is highly reproducible and cost-effective (U.S. $50 per case).