This paper analyzes the optimization of ambulance base locations in the city of Sorocaba, São Paulo, Brazil, through clustering methods and synthetic data generation. Sorocaba currently operates with a single SAMU ambulance base, located near the city center, which results in long travel distances and potentially delayed emergency responses. To address this issue, synthetic datasets of emergency calls were created to simulate incidents across the city, maintaining confidentiality while ensuring realistic data. Four clustering techniques—K-means, Gaussian Mixture Model (GMM), Agglomerative Clustering, and Spectral Clustering—were applied to identify alternative base configurations and compare them to the existing single-base system. The evaluation considered maximum and total distances traveled to respond to calls. Results showed that implementing multiple bases significantly improved system performance, reducing both total and maximum distances by more than 50% compared to the current setup. For instance, with four or more clusters, the total distance decreased from 11,084.75 km to less than 7,000 km, and the maximum distance dropped from 26.422 km to below 11 km, depending on the method used. Among the clustering approaches, K-means and GMM provided the most consistent balance between total and maximum distance reduction. These findings demonstrate the potential of clustering-based approaches to redesign emergency medical service systems and highlight the advantages of using synthetic data to overcome limitations in obtaining sensitive real-world datasets. The publicly available synthetic data generator developed in this study enables replication and adaptation to other urban contexts, providing a valuable tool for researchers and decision-makers. In addition, the paper recommends advancing the analysis by applying multi-criteria decision-making methods to define the optimal number of bases and discrete-event simulation to determine the appropriate number of ambulances required. The contributions of this research provide practical evidence that relocating ambulance bases using clustering methods can enhance emergency response efficiency, reduce travel distances, and potentially increase survival rates in urban populations.