In this study, articles in English and Chinese were selected from available electronic databases prior to September 2014. The metabolic concentrations and patterns of N-acetylaspartic acid (NAA), Choline (Cho), Creatine (Cr), NAA/Cr, NAA/Cho, and Cho/Cr ratios in radiotherapy-induced radiation encephalopathy by proton magnetic resonance spectroscopy were extracted. A meta-analysis was performed to quantitatively synthesize findings of these studies. Weighted mean difference (WMD) and 95% confidence intervals (95%CIs) were calculated using random or fixed effective models. Heterogeneity between studies was assessed using the Cochrane Q test and I (2) statistics. The results indicated that a total of 4 researches involving 214 patients met inclusion criteria. Depending on methodologies of selected studies, control groups were referred to as healthy subjects. The combined analysis revealed that there was no significant difference in value of Cr between radiotherapy group and healthy control group (WMD = -1.483, 95% CI: -67.185-64.219, p = 0.965). However, there were significant difference in values of NAA (WMD = -18.227, 95%CI: -36.317--0.137, p = 0.048), Cho (WMD = 38.003, 95%CI: 5.155-70.851, p = 0.023), NAA/Cr (WMD = -1.175, 95%CI: -1.563--0.787, p = 0.000), NAA/Cho (WMD = -1.108, 95%CI: -2.003-0.213, p = 0.015), and Cho/Cr (WMD = -0.773, 95%CI: 0.239-1.307, p = 0.005). In conclusion, MRS can be regarded as an effective and feasible imaging test for radiotherapy-induced radiation encephalopathy in NPC patients.