Abstract Background: Anorectal melanoma (AM) is generally staged using criteria developed and validated for cutaneous melanoma (CM), primarily due to its rarity and the lack of a specific staging system, despite limited evidence to support this. In an attempt to risk-stratify AM patients (pts), we analyzed the performance of the AJCC 8th edition CM staging system and a system recently developed for vulvar melanoma (VM) in a retrospective cohort. Methods: Demographics, clinicopathologic factors, and follow-up information were collected for 160 AM pts treated at our institution. The pts were grouped based on their clinical stage at presentation (localized to anorectum [L], regional metastases [R], distant metastases [D]). For further risk stratification, L pts were grouped according to the following systems: (i) AJCC 8th edition CM T-categories (T1a to T4b, 8 tiers); (ii) AJCC 8th edition CM stage (I vs. II, 2 tiers); (iii) the tumor thickness (TT) system (thin [T1], intermediate [T2-T3], thick [T4], 3 tiers); and TM system previously derived for VM (T1: TT≤ 2.0mm and mitotic rate (MR) <2/mm2 and T2: TT >2.0mm and MR <2/mm2). Univariate (UV) and multivariable (MV) Cox proportional hazards regression modeling determined associations with disease specific survival (DSS). The Kaplan-Meier method estimated overall survival (OS) and DSS. Results: The cohort (n=160) included 67 L, 55 R and 38 D pts. With median follow-up of 1.63 years (y), the median DSS was 1.75y for all pts. DSS progressively decreased according to stage at presentation (L: 2.39y; R: 1.81y and D: 1.25y). By UV analysis, clinical stage at presentation correlated with DSS (R: p=0.05, HR=1.52; D: p<0.001, HR=3.24, compared to L). By MV analysis (including clinical stage, tumor thickness, regression [REG], lymphovascular invasion [LVI], perineural invasion and resection margin status), the presence of distant metastasis correlated with DSS (p<0.001, HR=2.71), in addition to tumor thickness, REG, and LVI. To further optimize risk modeling and clinical management, we analyzed L pts using 4 distinct T-categorization systems. The 3-tier TT system robustly stratified DSS (p=0.03; HR- intermediate=2.36, thick=4.98, compared to thin). AJCC stage also classified 2 groups of pts according to DSS, with a trend towards significance (p=0.1; HR=3.37). In contrast, only OS models could be derived for the AJCC 8th edition T-categories (p=0.14) and TM system (p=0.1), with the latter yielding clear differences in outcomes, compared to the former. Analysis was also performed on a combined cohort of L pts with VM and AM, to determine if there are unifying themes among anogenital mucosal melanomas, and all the 4 systems stratified the pts (p<0.001), with stage, TT, and TM systems yielding clear differences in outcomes, compared to the 8 AJCC T-categories. Conclusion: Clinical stage at presentation and T-categorization of L pts based on modifications of the 8th edition AJCC CM staging criteria (stage and TT system) appear to be potentially informative tools to stratify risk in AM pts. Citation Format: Priyadharsini Nagarajan, JIn Piao, Jing Ning, Laura E. Noordenbos, Jonathan L. Curry, Carlos A. Torres-Cabala, A. Hafeez Diwan, Phyu P. Aung, Doina Ivan, Merrick I. Ross, Richard E. Royal, Jennifer A. Wargo, Wei-Lien Wang, Rashmi Samdani, Alexander J. Lazar, Asif Rashid, Michael A. Davies, Victor G. Prieto, Jeffrey E. Gershenwald, Michael T. Tetzlaff. Prognostic model for disease-specific survival in anorectal melanoma [abstract]. In: Proceedings of the AACR Special Conference on Melanoma: From Biology to Target; 2019 Jan 15-18; Houston, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(19 Suppl):Abstract nr A11.