A system has been developed for failure detection and identification in the depth and heading control of an AUV. A redundancy management technique was implemented using the CLIPS expert system shell. The term redundancy, as used here, does not mean that sensors are duplicated but that independent values of the same quantity can be calculated by combining data from several different sensors. The rules used for failure detection and identification are presented and discussed. This failure detection scheme was implemented and tested on the simulator for the Texas A&M AUV Controller. Failures were introduced and the performance of the system was evaluated based on its accuracy and time response in correctly detecting and identifying failures. All single failures and most multiple failures were detected and identified correctly. False alarms were avoided by requiring several successive occurrences of an aberration before it was recognized as a failure.