We propose Transformer-based Alzheimer’s disease (AD) analyzer for 3D MRI. The proposed network can analyze 3D MRI images efficiently combining 3D CNN, and Transformer. It is possible to efficiently extract locality information for AD-related abnormalities in local brain based on CNN networks with inductive bias. Also, the transformer network is also used to obtain attention relationship among 3D representation features after CNN. Our proposed method was compared to various networks including 3D CNN and transformer with an area under curve and accuracy for AD classification in multi-institutional datasets. Also, the transformer interpretability technique-based activation map can visualize AD-related abnormality region.