Are expert witnesses biased by the side (defense versus prosecution) that hires them? We examined this issue by having students act as expert witnesses in evaluating interviews in a child sexual abuse case (Experiment 1, N = 143) and tested the value of an instruction to counteract such allegiance effects. The intervention concerned an instruction to consider arguments both for and against the given hypothesis (i.e., two-sided instructions; Experiment 2, N = 139). In Experiment 3 (N = 123), we additionally provided participants with three different scenarios. Participants received a case file regarding a case of alleged sexual abuse. With the file, participants received an appointment letter emphasizing elements of the file that questioned (defense) or supported (prosecution) the veracity of the accusation. Participants displayed allegiance bias (Experiments 1-3), but two-sided instructions were not successful in eliminating allegiance bias (Experiments 2, 3). The findings underscore the importance of legal safeguards in expert witness work.