The purpose of this study was to determine a convenient method for the modification of architectural grade by nuclear features and to evaluate the prognostic significance of the new International Federation of Gynecology and Obstetrics (FIGO) grading system by studying 288 patients with endometrioid endometrial carcinoma. All patients were subjected to initial surgical exploration and staging by 1988 FIGO guidelines. Three different grading systems were evaluated for their prognostic value: architectural, nuclear, and FIGO combined systems. All three grading systems significantly predicted poor survival, but only the FIGO grade (p < 0.001), stage (p < 0.001), and cervical involvement (p = 0.04) remained significant in multivariate analysis. In the architectural grade 2 group, the 5-year survival rate for 39 patients with grade 1 or 2 nuclei was 87%, compared with 66% for 35 patients with grade 3 nuclei (p = 0.03). In the architectural grade I group, the 5-year survival rate for 84 patients with grade 2 nuclei was 93% without significant difference from the original group (96%). FIGO grade 3 tumor predicted 70% of deaths (29/41), whereas architectural grade 3 tumor detected 41% (17/41) of deaths (p = 0.001). In conclusion, in determining the FIGO grade, upgrading of architectural grade I or 2 tumors by grade 3 nuclei was the most reliable method. The new FIGO grading system was prognostically superior to the previously used architectural grading system.