Objectives: To study application of the maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV) and total lesion glycolysis (TLG) with $^{18}F$-FDG PET/CT for predicting prognosis of esophageal squamous cell cancer (ESC) patients. Methods: Eighty-six patients with ESC staged from I to IV were prospectively enrolled. Cisplatin-based chemoradiotherapy (CCRT) or palliative chemoradiotherapy were the main treatment methods and none received surgery. $^{18}F$-FDG PET/CT scans were performed before the treatment. SUVmax, MTV, and TLG were measured for the primary esophageal lesion and regional lymph nodes. Receiver operating characteristic curves (ROCs) were generated to calculate the P value of the predictive ability and the optimal threshold. Results: MTV and TLG proved to be good indexes in the prediction of outcome for the ESC patients. An MTV value of 15.6 ml and a TLG value of 183.5 were optimal threshold to predict the overall survival (OS). The areas under the curve (AUC) for MTV and TLG were 0.74 and 0.70, respectively. Kaplan-Meier analysis showed an MTV less than 15.6 ml and a TLG less than 183.5 to indicate good media survival time (p value <0.05). In the stage III-IV patient group, MTV could better predict the OS (P < 0.001), with a sensitivity and specificity of 0.80 and 0.67, respectively. Conclusions: Pre-treatment MTV and TLG are useful prognostic factors in nonsurgical ESC.