The management of Vocal Fold Leukoplakia (VFL) remains problematic. There is no consensus on the indications or the timing for surgery. The objective was to select the most accurate classification for predicting low- and high-risk VFL in White Light Imaging (WLI) and Narrow Band Imaging (NBI) and to establish a diagnostic algorithm with a timely referral for treatment. A total of 259 VFL patients were included in the study; 186 lesions were classified as low-grade and 110 as high-grade dysplasia. The results of WLI acc. to the two-tier and the three-tier Chen 2019 classifications and NBI classifications: ELS, Ni 2011, and Ni 2019 with different cut-off points were compared with the pathological examination (HP). In WLI, the greatest agreement was obtained between type 3 of the three-tier classification and high-grade dysplasia (accuracy, specificity, and PPV: 80.4%, 92.0%, and 81.5%, respectively). Assessing VFL periphery in NBI, cut-off point 5 (Ni 2011 type V) demonstrated a higher accuracy, specificity, and PPV than 4 (83.1%, 93.6%, 85.5% and 77.4%, 74.9%, and 65.4%, respectively). In NBI, we observed higher accuracy, sensitivity, and PPV (84.1%, 93.0%, 85.2% vs. 80.7%, 81.3% and 71.3%, respectively) for cut-off point 5 (Ni 2019 type V and VI) in comparison to the cut-off point 4 group (type IV, V, and VI) (80.7%, 81.3%, 71.3%, respectively), and a higher kappa value (0.68 vs. 0.58) was obtained. We have shown that both the plaque image and the microvascular pattern on the leukoplakia periphery are critical in the diagnosis of high-risk VFL. The most accurate predictor of VFL malignant transformation in WLI is type 3 according to the Chen 2019 classification, while in NBI type V and VI according to the Ni 2019 classification.