Shih, Feng-Hsien WarrenNakajima, KazuoWe show that the adaptive diagnosis approach can be applied to probabilistically diagnosable systems. Three adaptive diagnosis algorithms are developed under both the symmetric and asymmetric test invalidation assumptions. A test selection strategy based on a probabilistic measure of test results is derived and used in each algorithm. We show that the adaptive algorithms are efficient in identifying all faulty units in a system.en-USTechnical Report