Web-enabled animated agent technology has advanced to the point that it would be feasible to create self-administered interviewing systems that use them; for example, in the domain of intelligent tutoring systems the University of Memphis’ AutoTutor system, which engages in tutorial dialogue with students, leads to learning gains comparable to those with human tutors. But little is known about the benefits and drawbacks of using such systems for survey interviewing. The current study explores how an animated agent interviewer, built using the AutoTutor platform, affects answers to open-ended questions, relative to human interviewers, CASI, and instant messaging. In the study, seventy-eight college students participated in interviews in which they provided answers to questions about their beliefs about alcohol consumption, their alcohol consumption behaviors, their personal lives, and their family histories. The participants in the study were assigned to one of four conditions: (1) animated agent (AA), (2) text-only (CASI), (3) instant messaging (IM), and (4) human-to-human (FTF). The interview questions were the same in all four conditions and the length of the sessions ranged from approximately 10 minutes to 45 minutes. Results indicated that animated agents did not interfere with participants’ willingness to disclose personal information about their beliefs and behaviors regarding alcohol use; intriguingly, participants in the FTF condition used less semantically rich language and fewer emotion words than when interacting with the animated agent. One interpretation of these finding is that face-to-face interviews are face-threatening and that computer-based technologies (e.g., animated agents) may be a viable option for collecting personal and sensitive information in initial interviews. These along with other results from this study inform research on the role of social agency and human-like personas in interviews that require participants to disclose personal information.