Simple Summary: We aimed to identify risk factors and evaluate the accuracy of existing risk stratifications developed for non-muscle-invasive bladder cancer (NMIBC) regarding their ability to predict high-grade recurrence and progression. We included 171 NMIBC patients treated with TURBT and adjuvant BCG, of whom 73 experienced recurrence (42.7%), and 29 developed progression (17%). Available risk models (EORTC/ CUETO/ EAU) demonstrated limited accuracy in predicting high-grade recurrence-free (RFS) and progression-free survival (PFS). Multivariable analysis identified independent predictors for high-grade RFS, including T1HG tumor at repeat TURBT, tumor multiplicity, previous history of high-grade NMIBC, and EORTC2006 progression risk score. In conclusion, the available risk models lack accuracy in predicting high-grade RFS and PFS in BCG-treated NMIBC, suggesting potential improvement with the inclusion of additional risk factors. The currently available EORTC, CUETO and EAU2021 risk stratifications were originally developed to predict recurrence and progression in non-muscle-invasive bladder cancer (NMIBC). However, they have not been validated to differentiate between high-grade (HG) and low-grade (LG) recurrence-free survival (RFS), which are distinct events with specific implications. We aimed to evaluate the accuracy of available risk models and identify additional risk factors for HG RFS and PFS among NMIBC patients treated with Bacillus Calmette–Guérin (BCG). We retrospectively included 171 patients who underwent transurethral resection of the bladder tumor (TURBT), of whom 73 patients (42.7%) experienced recurrence and 29 (17%) developed progression. Initially, there were 21 low-grade and 52 high-grade recurrences. EORTC2006, EORTC2016 and CUETO recurrence scoring systems lacked accuracy in the prediction of HG RFS (C-index 0.63/0.55/0.59, respectively). EAU2021 risk stratification, EORTC2006, EORTC2016, and CUETO progression scoring systems demonstrated low to moderate accuracy (C-index 0.59/0.68/0.65/0.65) in the prediction of PFS. In the multivariable analysis, T1HG at repeat TURBT (HR = 3.17 p < 0.01), tumor multiplicity (HR = 2.07 p < 0.05), previous history of HG NMIBC (HR = 2.37 p = 0.06) and EORTC2006 progression risk score (HR = 1.1 p < 0.01) were independent predictors for HG RFS. To conclude, available risk models lack accuracy in predicting HG RFS and PFS in -NMIBC patients treated with BCG. [ABSTRACT FROM AUTHOR]