Age-related macular degeneration (AMD) is a disease that results in distortion and blurriness in the center of vision and is the world's fourth leading cause of blindness. The disease is caused by retinal degeneration due to age and can be diagnosed through the fundus in ophthalmology. Since AMD sufferers do not experience the initial symptoms, early diagnosis by ophthalmology and routine checks are needed to prevent the disease from worsening without any treatment. Machine learning has been applied in diagnosing disease, and one of such approaches that have generated good results is Support Vector Machines (SVM). Deep learning, as a branch of machine learning, has also been used for diagnosis, especially imaging with the Convolutional Neural Network (CNN) method, which has given good results as well. Therefore, CNN and SVM were combined in this study to solve the classification problem of AMD based on fundus images. CNN was used for feature extraction, and the results were employed by SVM for the classification. SVM used three common kernel functions, namely linear, polynomial, and radial basis function (RBF). The performance evaluation applied the hold-out validation technique, as well as metric accuracy, precision, and recall, and the results revealed that CNN-SVM with RBF kernel had the highest accuracy and recall but the least precision. Consequently, this method is recommended for future research on disease diagnosis.