This article shifts the focus of prior research examining virtual work to investigate how supervisors who work virtually use subordinate performance information. Drawing insights from several research streams, in Study 1, we propose that supervisors who work virtually bias performance ratings in the direction of information that is observed directly in the office, rather than that which is received when working virtually. In Study 2, we replicate and extend these results to show that this bias is independent of the level of performance information received. Results also indicate that, for high-performing workers, performance information received virtually is evaluated more extremely than information observed directly. We did not find evidence for this extremity effect when low-performing workers were evaluated.