Brain tumor detection is a critical component of medical diagnostics, aiming to provide accurate, timely identification of tumor presence. This study utilizes a Convolutional Neural Network (CNN) approach with the VGG-16 model architecture to classify brain MRI images as either showing the presence of a tumor or not. Leveraging transfer learning, VGG-16 was fine-tuned for binary classification on a dataset of brain MRI images. This approach achieved validation and test accuracies of approximately 88% and 80%, respectively. Our methodology combines image preprocessing techniques with data augmentation to enhance model robustness on limited datasets. The results demonstrate the potential of CNN-based deep learning models in automated medical imaging and suggest future improvements through dataset expansion and model fine-tuning.