Staging of renal pelvic urothelial carcinoma can be challenging due to anatomic variation at the renal pelvis compared with ureter and bladder and calls into question the prognostic accuracy of the current TNM staging. In this study, we determined staging and cancer-specific survival (CSS) in 141 patients undergoing nephroureterectomy for renal pelvic urothelial carcinoma (pTa=50, pT1=29, pT2=10, pT3=36, and pT4=16). Under current staging criteria, we found no significant difference in CSS between adjacent staging categories step-wise across pTa, pT1, pT2, and pT3 tumors. When pT3 tumors were subcategorized into renal medulla, peripelvic adipose, or renal cortex invasion with or without peripelvic adipose invasion, we found that cortical invasion was associated with significantly worse CSS compared with medulla or peripelvic adipose invasion only. We next revised staging criteria such that pT1 correlated with invasion of lamina or muscularis propria (n=37), T2 with invasion of medulla or peripelvic adipose only (n=26), and pT3 with cortical invasion (n=12). Under the new criteria, better separation of survival curves was achieved; however, pT1 and pT2 remained statistically insignificant. When further redefining pT3 as invasive of cortex only (n=12) and combining medulla with lamina and muscularis propria invasion as a lower stage (pT1, n=63), there was further improvement in the prognostic stratification. Therefore, our data show that consideration of revised and simplified T staging criteria at the renal pelvis is warranted, wherein invasion of any anatomic structure up to the cortex shows a similar prognosis (combined pT1 category) and invasion of cortex showing significantly worse prognosis (pT3).