This study aims to enhance fingerprint classification accuracy by optimizing the K value in the K-Nearest Neighbor (KNN) algorithm using Particle Swarm Optimization (PSO). Accurate fingerprint labeling is crucial for security and access control applications, with KNN being an effective machine learning algorithm for classifying unlabeled data. However, determining the optimal K value in KNN is a challenge, as suboptimal K values can lead to inaccurate classification results. Optimization of the K value is performed using PSO, a population-based search algorithm that initializes the population randomly with particles or individuals. The research findings indicate that using PSO for KNN optimization yields significantly better results compared to conventional methods for setting the initial K value. Specifically, PSO determined an optimal K of 3, and with an 80:20 dataset ratio, achieved an accuracy of 82.5%. Testing with this ratio was conducted 39 times to ensure result consistency. However, accuracy calculated using the confusion matrix showed a result of 96.15%, which may be influenced by data distribution during testing. Therefore, optimizing the initial K value using PSO in KNN can improve fingerprint recognition classification accuracy.