The brain tumor is regarded as a serious cancerous tumor that if not detected and accurately identified, may lead in the patient's death. Therefore, recent advances in the field of deep learning (DL) have assisted radiologists in diagnosing tumors with high accuracy and speed when compared to manual diagnosis, which requires the radiologist's effort and competence. Oncologists typically perform the initial evaluation of brain tumors using medical imaging techniques such as computerized tomography (CT) and magnetic resonance imaging (MRI). These two medical imaging techniques are commonly used to create highly detailed images of the brain's structure to monitor any changes. A surgical biopsy of the suspected tissue (tumor) is required for a detailed diagnosis by the specialist if the doctor suspects a brain tumor and needs more information about its type. These various techniques in brain tissue imaging have increased image contrast and resolution in recent years, allowing the radiologist to identify even small lesions and thus achieve higher diagnostic accuracy. This research introduced an automatic system using a Convolutional Neural Network (CNN) to classify MRI brain tumor images consisting of various layers, and then selected the best system that achieved an accuracy of 99.6% with different images sizes and learning rates.