Fault diagnosis methods for specified fault models might deduce wrong faults as suspicious candidate faults (misprediction). The methods might not be able to also deduce suspicious candidate faults (non-prediction). In this paper, a fault diagnosis method for a single universal logical fault model in scan testing is proposed. In the fault diagnosis method, a diagnostic fault simulation for a single universal logical fault model is used. The number of mispredictions, the number of non-predictions, the number of suspicious candidate faults, and fault diagnosis time are evaluated compared with a fault diagnosis method for a single stuck-at fault model. Experimental results show the effectiveness of our proposed method.