In-service and preservice teachers are increasingly required to integrate research results into their classroom practice. However, due to their limited methodological background knowledge, they often cannot evaluate scientific evidence firsthand and instead must trust the sources on which they rely. In two experimental studies, we investigated the amount of this so-called epistemic trustworthiness (dimensions expertise, integrity, and benevolence) that student-teachers ascribe to the authors of texts who present classical research findings (e.g., learning with worked-out examples) that allegedly were written by a practitioner, an expert, or a scientist. Results from the first exploratory study suggest that student-teachers view scientists as "smart but evil," since they rate them as having substantially more expertise than practitioners, while also being less benevolent and lacking in integrity. Moreover, results from the exploratory study suggest that evaluativistic epistemic beliefs (beliefs about the nature of knowledge) predict epistemic trustworthiness. A preregistered conceptual replication study (Study 2) provided more evidence for the "smart but evil" stereotype. Further directions of research as well as implications for practice are discussed.