Providing models for model-based diagnosis has always been a challenging task. There has never been an agreement on an underlying modeling language, making it almost impossible to share models within our community. In addition, there are other domains like formal methods or model-based testing relying on system models for formal verification and automated test case generation. Although, there we face the situation of different modeling languages as well, the question remains whether it is possible to re-use these models in the context of model-based diagnosis. In this paper , we elaborate on this question and show how models written in LNT can be used for fault local-ization only requiring simple modification. This allows re-using formal method's models for diagnosis directly. Besides discussing the underlying principles, we also present a use case showing the applicability of the methods.