Role-playing using avatars has been demonstrated as effective, but it has not been compared to role-playing with a live actor. In this study, undergraduate psychology students (N=93) conducted a clinical interview in a role-play with an avatar or live actor. Mixed methods included a survey and written reflection. The study found no differences in student interest-enjoyment, skills-knowledge, comfort level, or utility value in survey scales found to be reliable. High levels of active learning were found in qualitative data in both conditions, though it was significantly higher in the avatar group. High levels of benefits from group work also were found. An awkwardness, or uncanny valley effect, came up, though significantly more in the avatar group. Implications for teaching include using either avatar- or actor-based role-playing depending on availability, using a group format, and minimizing student awkwardness.