A concept to improve the resilience to failures of the Cambridge Ring network is described. The ideas conceived should be applicable to any other ring network. Reliability issues are raised, and several existing fault-tolerant techniques are briefly discussed. The Centre-Switching Ring- Star concept is then described, detailing its design, operation and problems. The star connections are only brought into use when a fault is detected. This design is based on the automatic bypassing of faulty repeaters or cable sections. Any number of failures may be tolerated. The design allows the continual operation of the network whilst a new node is being added or removed. A laboratory model has been built and successfully tested.