Background: Urachal carcinoma (UrC) is a rare, aggressive cancer with a poor prognosis that is frequently diagnosed in advanced stages. Due to its rarity, the current staging systems, namely Sheldon and Mayo, were established based on relatively small patient cohorts, necessitating further validation. We used a large patient population from the National Cancer Database (NCDB) to model a novel staging system based on the Tumor (T), Node(N) and Metastasis (M) (TNM) staging system and compared it to established staging systems. Methods: Patients diagnosed with UrC between 2004 and 2016 were identified and included in the analysis using the NCDB. The proposed TNM staging system is defined as such: stage 1: T0, a, IS, or 1 and N0M0; stage 2: T2, T3, T3a and N0M0; stage 3: T3b, T4a and N0M0; stage 4: T4b and N0M0, or any T and/or nodal and/or distant metastasis. The Sheldon and Mayo staging systems were correlated with the TNM staging system. Kaplan-Meier (KM) survival plots were generated and stratified by stage to assess the median overall survival (OS) of corresponding stages of each staging system. A cox proportional-hazards regression model was developed to highlight predictors of overall survival. Results: A total of 626 patients with UrC were identified with a median age of 58, of whom 56% (n = 350) were males. The OS for the entire cohort was 58.2 (50.1-67.8) months (mo) with survival rates of 83%, 70%, and 49% at 12, 24, and 60 mo, respectively (p < 0.0001). With the Sheldon staging system, the median OS was 132.2 (89.2-132.2) for stage 1, undefined for stage 2 due to median OS not reaching 50%, 69.6 (60.8-89.5) for stage 3, and 17.3 (14.2-20.7) mo for stage 4 (p < 0.0001 for all stages). With the Mayo staging system, the median OS was 132.2 (88.6-132.2), 69.5 (58.5-88.4), 19.7 (11.6-32.0), and 16.8 (13.2-20.7) mo for stage 1, 2, 3, and 4, respectively (p < 0.0001). The median OS based on the TNM staging system was 132.2 (88.6-132.2), 64.6 (51.8-83.0), 42.7 (30.1-58.2), and 17.8 (11.0-21.1) mo for stage 1, 2, 3, and 4, respectively (p < 0.0001). Based on the Sheldon system, most patients were staged as stage 3 (n = 349; 56%) and 4 (n = 177; 28%) and KM survival curves overlapped between stages 1 and 2. Per the Mayo system, most patients were staged as stage 1 (n = 209; 33%) and 2 (n = 240; 38%) and KM survival curves overlapped between stages 3 and 4. In comparison, our TNM staging system had a more balanced sample and survival distribution per stage and no overlap among stages on KM survival curves. The Mayo and TNM staging systems were more accurate in terms of stage-survival correlation than the Sheldon staging system (p < 0.05 for all stages). Conclusions: The proposed novel TNM staging system for UrC has a more balanced sample distribution and a more accurate stage-survival correlation than the traditional Mayo or Sheldon staging systems. It is clinically applicable and enables better risk stratification, prognosis, and therapeutic decision-making.