Purpose of investigation: To investigate the performance of MRI and/or gross examination of specimen in detecting cervical invasion and factors that may influence MRI’s performance. Materials and Methods: Endometrial cancer patients who had hysterectomy in Prince of Wales Hospital in Hong Kong from January 2007 to November 2014 were identified retrospectively. Those who had preoperative MRI assessment for cervical invasion were included. Patient’s records were reviewed for demographic, operative, MRI, and pathological findings. The accuracy of MRI and operative findings were determined by correlating with pathological findings. Results: A total of 318 patients were included. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio of MRI in diagnosing cervical invasion were 86.5%, 48%, 93.7%, 58.5%, 90.6%, 7.57, and 0.56, respectively. Area under the curve was 0.71. The accuracy, sensitivity, and specificity of gross examination were 89.6%, 51.9% and 96.4%, respectively. The accuracy, sensitivity and specificity of combined assessment with MRI and gross examination were 85.8%, 56.3%, and 91.2%, respectively. Conclusions: MRI has high accuracy and specificity but low sensitivity in detecting cervical invasion. The performance of gross examination of specimen is comparable to MRI. MRI cannot be recommended as sole assessment method to detect cervical invasion by endometrial cancer and a higher sensitivity can be achieved by the combined assessment approach.