Early and accurate diagnosis of brain tumors is essential for effective treatment and improved patient survival. Although deep learning models achieve high performance in MRI interpretation, their limited transparency constrains clinical adoption. This paper proposed an explainable artificial intelligence (XAI) system for brain tumor diagnosis using magnetic resonance imaging. The proposed framework leverages a pretrained model for tumor classification and integrates Grad-CAM to generate interpretable visual explanations. Experimental results demonstrate a classification accuracy of up to 99%, with results consistent with clinical expertise, highlighting the potential of the system as a reliable decision support tool in medical imaging.