The transition from manual to automatic cephalometric landmark identification has not yet reached a consensus for clinical application in orthodontic diagnosis. The present umbrella review aimed to assess artificial intelligence (AI) performance in automatic 2D and 3D cephalometric landmark identification.