Abstract: The present application discloses a server load balancing method and apparatus, and a server device. The method includes: classifying, according to link quality of back-end servers in a server cluster, the back-end servers into multiple server groups; setting priorities of the multiple server groups according to link quality of the server groups; selecting, when an access request sent by a first terminal is received, a server group from the multiple server groups as a target server group based on the priorities; selecting a target back-end server from the target server group; and establishing communication between the target back-end server and the first terminal.