This study aimed to design automated volumetric-modulated arc therapy (VMAT) plans in Pinnacle auto-planning and compare it with manual plans for patients with lower thoracic esophageal cancer (EC). Thirty patients with lower thoracic EC were randomly selected for replanning VMAT plans using auto-planning in Pinnacle treatment planning system (TPS) version 9.10. Historical plans of these patients were then compared. Dose-volume histogram (DVH) statistics, dose uniformity, and dose homogeneity were analyzed to evaluate treatment plans. Auto-planning was superior in terms of conformity index (CI) and homogeneity index (HI) for planning target volume (PTV), significantly improving 8.2% (p = 0.013) and 25% (p = 0.007) compared with manual planning, respectively, and decreasing dose of heart and liver irradiated by 20 to 40 Gy and 5 to 30 Gy, respectively (p < 0.05). Meanwhile, auto-planning further reduced the maximum dose (D-max) of spinal cord by 6.9 Gy compared with manual planning (p = 0.000). Additionally, manual planning showed the significantly lower low-dose volume (V-5) for the lung (p = 0.005). For auto-planning, the V-5 of the lung was significantly associated with the relative volume index (the volume ratio of PTV to the lung), and the correlation coefficient (R) and p-value were 0.994 and 0.000. Pinnacle auto planning achieved superior target conformity and homogeneity and similar target coverage compared with historical manual planning. Most of organs at risk (OARs) sparing was significantly improved by auto-planning except for the V-5 of the lung, and the low dose distribution was highly associated with PTV volume and lung volume in auto-planning. (C) 2017 American Association of Medical Dosimetrists.