11164 Background: tMNs are rare, albeit serious complication in cancer survivors. A tMN risk prediction model is currently not available, therefore we leveraged clinical information and history in a large population-claims database to devise the first tMN risk prediction model in adult cancer survivors. Methods: SEER-Medicare was used to analyze 970,390 adults aged 66-84 diagnosed with first primary cancer (FPC) from 2000-2011 (with follow-up through 2015), who survived ≥1 year. Medicare claims data classified individuals with a priori identified conditions. We focused on prostate, GI, breast, lung, and bladder (n=830,468) FPCs to develop a tMN risk prediction model. The population was divided into training and validation cohorts. Χ2 test identified risk factors associated with tMN. The tMN risk score (TMNRS) was created as a simple arithmetic sum of independent predictors of tMN weighted according to the adjusted odds ratio from logistic regression analysis. Pts were categorized into risk groups (Table) based on their TMNRS, which were tested in the validation cohort. Model performance was evaluated using ROC and c-statistics. Results: 87-96% of tMN developed within the first 10 years of FPC. Predictors of tMN included history of autoimmunity, infections, cardiovascular disease, granulocyte-colony stimulating factor, FPC stage, chemoradiotherapy exposure, age at FPC diagnosis, and sex. Training analysis distinguished survivors into distinct risk groups (table) with 10-year incidence of tMN ranging from 0.3% to 1.5% (prostate), 0.04% to 1.1% (GI), 0.2% to 1.4% (breast), 0.1% to 1.2% (lung) and 0.1% to 1.4% (bladder). The table summarizes the incidence of tMN and by risk category for each of the FPCs. After validation, the c-statistic ranged from 0.62-0.68. Conclusions: We utilized the largest SEER-Medicare cohort to date to develop a tMN risk model. The TMNRS captures several novel predictors and can be readily used in the clinic. This easy-to-use tool distinguishes cancer survivors into clinically relevant groups at risk of tMN development. This study lays the framework for screening and monitoring of pts at high risk of tMN. In future, we will leverage increasingly available molecular data to improve prediction performance of the model. Web-based calculator to be published with the manuscript/available at time of presentation. [Table: see text]