OBJECTIVE: To investigate the accuracy of magnetic resonance imaging (MRI) in classifying sonographically indeterminate ovarian and adnexal masses. MATERIALS AND METHODS: This was a retrospective cross-sectional study of the unenhanced pelvic MRI scans of 243 patients with a collective total of 336 adnexal and ovarian masses. RESULTS: Unenhanced MRI showed a sensitivity of 97.7%, a specificity of 86.4%, and an accuracy of 93.8%. The area under the ROC curve was 0.944 (95% CI: 0.913-0.974). CONCLUSION: Our results show that an unenhanced MRI protocol can be used to classify adnexal masses, especially in clinical settings in which the intravenous administration of gadolinium-based contrast is not safe and should be avoided.