Abstract Objective Uterine sarcomas are aggressive cancers, often not recognized prior to surgical exploration. The goal of this study was to determine the utility of endometrial sampling in detecting uterine sarcomas and to examine factors associated with diagnostic inaccuracy. Methods All uterine tumors identified at hysterectomy from 1990 to 2006 were reviewed. Included patients underwent preoperative endometrial sampling reviewed at our center. Pathologic data was documented through review of hospital records. Statistical analysis was performed using Chi square test. Results 938 patients were identified. Preoperative sampling was available for review in 730 (78%) subjects. Uterine sarcomas occurred in 142 patients; 72 (51%) underwent preoperative sampling. Overall, endometrial sampling identified an invasive tumor in 84% (600/713), and the correct histology in 79% (564/713). Among women with sarcomas, preoperative sampling suggested an invasive tumor in 86% (62/72) and predicted the correct histologic diagnosis in 64% (46/72). The rate of detection of an invasive cancer by preoperative sampling was not statistically different among sarcomas and epithelial tumors (86% vs. 84%, p = 0.76). Preoperative sampling was significantly less reliable in predicting the correct histology for uterine sarcomas (64% vs. 81%, p < 0.0001). Similar trends were seen when sarcoma patients were compared to low-grade and high-grade epithelial cancers. Both biopsy and curettage had similar accuracy in diagnosing sarcomas ( p = 0.84). Conclusions Endometrial sampling has a significantly lower predictive value in diagnosing uterine sarcomas compared to epithelial uterine malignancies. Biopsy and curettage have similar accuracy. Novel diagnostic techniques are needed to accurately identify uterine sarcomas preoperatively.