Purpose: To assess efficacy of Ki67 combined with VEGF as a molecular grading model to predict outcomes with non-muscle invasive bladder cancer (NMIBC). Materials: 72 NMIBC patients who underwent transurethral resection (TUR) followed by routine intravesical instillations were retrospectively analyzed in this study. Univariate and multivariate analyses were performed to confirm the prognostic values of the Ki67 labeling index (LI) and VEGF scoring for tumor recurrence and progression. Results: The novel molecular grading model for NMIBC contained three molecular grades including mG1 (Ki67 $LI{leq}25%$, VEGF $scoring{leq}8$), mG2 (Ki67 LI>25%, VEGF $scoring{leq}8$; or Ki67 $LI{leq}25%$, VEGF scoring > 8), and mG3 (Ki67 LI > 25%, VEGF scoring > 8), which can indicate favorable, intermediate and poor prognosis, respectively. Conclusions: The described novel molecular grading model utilizing Ki67 LI and VEGF scoring is helpful to effectively and accurately predict outcomes and optimize personal therapy.