Abstract Objective In patients with endometrial cancer, obesity is associated with favorable prognostic characteristics but not with prolonged survival. The aim of this study was to elucidate the reason for this clinical paradox. Methods We retrospectively reviewed 1173 patients with endometrial cancer. Patients were divided into a non-obese group body mass index (BMI) < 30 kg/m2, class I obesity group (BMI 30–35 kg/m2) and class II obesity group (BMI ≥ 35 kg/m2). The relationship between clinicopathological factors and disease-specific survival (DSS) was analyzed by Cox regression analysis. To correct for three-time significance testing, we used the Bonferroni method, giving the level of probability at which findings were considered significant as P < 0.0167. Results Three disease-intrinsic variables—older age, advanced stage and high-risk histology—and three treatment-related variables—no hysterectomy, no lymphadenectomy and no chemotherapy—were independently associated with poor DSS. DSS was similar among the three groups of patients even though the proportion of patients with plural pretreatment-related unfavorable risk factors significantly decreased with increment of BMI category (40.1 vs. 27.5 vs. 17.6%, P = 0.0003). The proportion of patients with plural treatment-related unfavorable prognostic factors significantly increased with increment of BMI category (21.3 vs. 26.7 vs. 39.3%, P = 0.0072). Conclusions Poor-quality surgical staging in obese women may result in worse than expected survival outcomes. In patients with endometrial cancer, obesity is associated with disease-intrinsic, favorable prognostic characteristics. Lack of survival advantage in obese women may be attributable to poor-quality surgical staging.