The Silhouette Score measures how similar a sample is to its own cluster compared to other clusters. A higher score indicates well-separated clusters. It's often used as a metric to evaluate and compare clustering results without ground truth labels.