Abstract Purpose: To evaluate the extent to which magnetic resonance spectroscopy (MRS) lipid metabolites are accurate in predicting high-grade cervical cancer. Material and methods: This prospective single-centre pilot study included 20 cases with pathologically proven cervical cancer. They underwent pelvic magnetic resonance imaging (MRI) with MRS. Two radiologists, blinded to the histopathological results, with 10 years of experience in gynaecological imaging, independently analysed the MRI images and MRS curves, and a third one resolved any disagreement. Using the histopathological results as a standard test, the receiver operating characteristics (ROC) curve was utilised to calculate the optimal lipid peak (1.3 ppm) cutoff for predicting high-grade cervical cancer. The difference in MRS metabolites between low- and high-grade cervical cancer groups was estimated using the Mann-Whitney test. Results: The study included 11 high-grade and nine low-grade cervical cancer cases based on the histopathological evaluation. A lipid (1.3 ppm) peak of 29.9 was the optimal cutoff for predicting high-grade cervical cancer with 100% sensitivity, 77.8%, specificity, and 90% accuracy. Moreover, there was a significant difference between low- and highgrade cervical cancer cases concerning lipid peak at 0.9 ppm, lipid peak at 1.3 ppm, and the peak of choline with (p-value 0.025, 0.001, and 0.023), respectively. Conclusions: MRS might be considered a useful imaging technique for assessing the grade of cervical cancer and improving the planning of treatment. It shows a good diagnostic accuracy. Therefore, it can be adopted in clinical practice for better patient outcome. Key words: cervical cancer, lipid, choline, diagnostic accuracy.