Endometrial cancer is the most common gynecologic malignancy in the United States and both the incidence and associated mortality are rising. When detected early, endometrial cancer survival rates are dramatically improved. Despite this, no screening methods have been developed which can effectively detect either pre-malignant lesions or early-stage cancers. We conducted a prospective study to analyze the effectiveness of molecular analysis of uterine lavage fluid, taken at the time of evaluation for a definitive tissue-based diagnosis, to assess the use of targeted next generation sequencing (NGS) for detecting endometrial carcinomas. The women in this study were primarily either experiencing abnormal uterine bleeding or had abnormal pelvic ultrasound findings and were being evaluated by hysteroscopy and curettage for a tissue diagnosis. We detected somatic mutations in all women who were later diagnosed with endometrial cancer, even microscopic stage IA, grade 1. In addition, we also identified that half of the women in our study who did not have clinical evidence of cancer, nonetheless possessed a significant landscape of driver mutations. Our findings therefore suggest the apparently inimicable possibilities of a genomics- based approach for endometrial cancer screening and the discovery of prevalent driver mutations in clinically-defined non-cancerous tissue.