A brain tumor refers to the abnormal and uncontrolled growth of cells in certain brain tissues, in which the malignant types have a high mortality rate. Therefore, accurate diagnosis requires precise analysis, as even minor human error could lead to severe consequences. Magnetic resonance imaging (MRI) serves as a non-invasive diagnostic tool for detecting brain tumors. Hence, Computer-aided diagnosis (CAD) systems can enable early diagnosis and increase survival rates, thus minimizing reliance on expert radiological analysis of MRI scans. Over the course of three months, a dataset of over 700 brain MRI scans was curated and sorted into five distinct categories: low-grade glioma, high-grade glioma, meningioma, mimic, and normal brains. Convolutional neural networks (CNNs) have proven highly effective for detecting tumors in brain MRIs. This study utilized transfer learning with pre-trained VGG16 and VGG19 models to classify the presented dataset. The MRI scans were resized, several preprocessing techniques were tested, and data augmentation techniques were applied. The experimental results found that the VGG16 and VGG19 models achieved accuracy rates of 98% and 96%, respectively. Overall, our findings indicate that the proposed multi-class classification models can enhance clinical research and MRI classification efficiency.