Over the past decade, information technology has dramatically changed the context in which economic transactions take place. Increasingly, transactions are computer-mediated, so that, relative to human¬human interactions, human-computer interactions are gaining in relevance. Computer-mediated transactions, and in particular those related to the Internet, increase perceptions of uncertainty. Therefore, trust becomes a crucial factor in the reduction of these perceptions. To investigate this important construct, we studied individual trust behavior and the underlying brain mechanisms through a multi-round trust game. Participants acted in the role of an investor, playing against both humans and avatars. The behavioral results show that participants trusted avatars to a similar degree as they trusted humans. Participants also revealed similarity in learning an interaction partner's trustworthiness, independent of whether the partner was human or avatar. However, the neuroimaging findings revealed differential responses within the brain network that is associated with theory of mind (mentalizing) depending on the interaction partner. Based on these results, the major conclusion of our study is that, in a situation of a computer with human-like characteristics (avatar), trust behavior in human-computer interaction resembles that of human-human interaction. On a deeper neurobiological level, our study reveals that thinking about an interaction partner's trustworthiness activates the mentalizing network more strongly if the trustee is a human rather than an avatar. We discuss implications of these findings for future research.