Objectives: To evaluate the accuracy of magnetic resonance imaging (MRI) in predicting deep myometrial invasion, cervical stromal invasion, and pelvic lymph node involvement in the preoperative assessment of women with endometrial cancer. Methods: Patients with endometrial cancer, having preoperative MRI available, and hysterectomy performed in a regional hospital in Hong Kong between 1 January 2010 and 31 December 2014 were included. Those treated with neoadjuvant therapy or without staging MRI were excluded. Primary outcome measure was the correlation between deep myometrial invasion suggested by preoperative MRI and subsequent histopathology of the hysterectomy specimen. Imaging-pathological correlation of cervical stromal invasion and pelvic node involvement was also assessed. Results: Overall 90 women met the criteria and were included in this study. Sensitivity, specificity, and accuracy of preoperative MRI was respectively 83.3%, 88.9%, and 87.8% for predicting deep myometrial invasion; 22.2%, 98.6%, and 83.3% for predicting cervical stromal invasion; and 60.0%, 96.6%, and 91.2% for predicting pelvic node disease. Conclusions: Preoperative MRI is highly accurate in the detection of deep myometrial invasion, which is the single most important prognostic factor. It is moderately accurate and highly specific in the diagnosis of cervical stromal invasion and pelvic node metastasis. The favourable performance of preoperative MRI allows better surgical planning that may then translate into better patient outcome. Although not formally included in the FIGO staging system, it is a highly valuable adjunct in preoperative assessment of women with endometrial cancer.