Objective To investigate the value of in vivo proton magnetic resonance spectroscopy (MRS) in the assessment of hepatocellular carcinoma (HCC) and monitor its metabolic change shortly after transcatheter arterial chemoembolization (TACE). Methods In this prospective study, 28 consecutive patients with large HCC (≥3 cm in diameter) confirmed by fine needle aspiration biopsy were recruited. The ~ 1 H MRS of all hepatic lesions and some uninvolved liver parenchyma were performed with 1.5T whole body MR scanner. Among them, 15 cases were evaluated again about one week after TACE. The main metabolites such as choline and lipid before and after interventional therapy were measured to assess the early response of the tumor. Results The technical success rate of ~ 1 H MRS in liver was high (33/41, 80%), closely related to breath motion, location of lesion, and size of voxel. In spectra, the choline compound peak of HCC elevated compared with uninvolved liver parenchyma. After TACE, both the amplitude and the area of choline resonance peak significantly descended (choline-to-lipid ratios from 0.352±0.080 to 0.167±0.030, P=0.026; from 0.205±0.060 to 0.070±0.020, P=0.042, respectively); yet lipid resonance peak ascended. Conclusions In vivo ~ 1 H MRS is technically feasible for the evaluation of large focal hepatic lesions, however, the reproducibility and stability are not as good as routine MR scan. ~ 1 H MRS can monitor the early stage metabolic changes of HCC after TACE but limitation like quantification still exists. Objective To investigate the value of in vivo proton magnetic resonance spectroscopy (MRS) in the assessment of hepatocellular carcinoma (HCC) and monitor its metabolic change shortly after transcatheter arterial chemoembolization (TACE). Methods In this prospective study, 28 consecutive patients with large HCC (≧ 3 cm in diameter) confirmed by fine needle aspiration biopsy were recruited. The ~ 1 H MRS of all hepatic lesions and some uninvolved liver parenchyma were performed with 1.5 T whole body MR scanner. Among them, 15 cases were evaluated again about one week after TACE. The main metabolites such as choline and lipid before and after interventional therapy were measured to assess the early response of the tumor. Results The technical success rate of ~ 1 H MRS in liver was high (33/41, 80%) , closely related to breath motion, location of lesion, and size of voxel. In spectra, the choline compound peak of HCC elevated compared with uninvolved liver parenchyma. After TACE, both the amplitude and the area of ​​choline resonance peak significantly descended (choline-to-lipid ratios from 0.352 ± 0.080 to 0.167 ± 0.030, P = 0.026; from 0.205 ± 0.060 to 0.070 ± 0.020, P = 0.042, respectively) Conclusions In vivo ~ 1H MRS is technically feasible for the evaluation of large focal hepatic lesions, however, the reproducibility and stability are not as good as routine MR scan. ~ 1H MRS can monitor the early stage metabolic changes of HCC after TACE but limitation like quantification still exists.