Brain tumors arise from the abnormal proliferation of cells in the brain, necessitating early and precise detection to avert potentially life-threatening outcomes. Magnetic Resonance Imaging (MRI) is a widely employed diagnostic tool thanks to its capability to deliver detailed visuals of brain irregularities. Although radiologists typically interpret MRI scans, the growing volume of imaging data necessitates automated methods to boost accuracy and efficiency. Deep learning presents a solution by automating the evaluation of medical images, eliminating the need for manual feature extraction. Pre-trained neural networks, which leverage previously gained knowledge from extensive datasets, can enhance performance in scenarios with limited data or restricted computational resources. This study proposes a new strategy for classifying brain tumors utilizing the EfficientNetB1 model, which is fine-tuned with additional layers. The approach involves preprocessing MRI images through cropping and employing data augmentation techniques such as rotation, height shifting, and horizontal flipping to improve the model’s generalization capabilities. The model was trained to differentiate between the three most prevalent types of brain tumors—Glioma, Meningioma, and Pituitary—along with a category for “no tumor.” The findings indicate a training accuracy of 97.67% and a validation accuracy of 89.48%, underscoring the proposed model’s effectiveness and reliability.