Skin cancer requires early diagnosis and appropriate treatment, particularly in distinguishing between benign and malignant tumors. This study proposes a two-stage classification method based on Vision Transformer (ViT) to classify skin cancer into four categories using skin lesion images. In the first stage, we perform binary classification to differentiate between benign and malignant <br/>lesions. If a lesion is classified as malignant, we further classify it into BCC (Basal Cell Carcinoma), SCC (Squamous Cell Carcinoma), or Melanoma in the second stage. This approach provides higher classification accuracy and stability compared to a single 4-class model, as specialized models for each stage yield more refined results. The study utilized 10,068 skin cancer images collected from March 29, 2004, to January 5, 2024, at Kyungpook National University Hospital and other medical centers.