It has been pointed out that conventional experimental procedures might distort results about lie recognition. Levine (2014) argues in his Truth–Default Theory (TDT) that people usually trust others but that they can quickly abandon their trust when witnessing a trigger event in which they examine why the person told a lie. He also stated that asking participants to judge which stimulus persons are lying in experiments can work as a “lie prime” (i.e., a trigger event). Furthermore, when participants are asked to think about lies, this may increase the estimated frequency of lies. We conducted two studies to investigate the effects of the following instruction: “Please determine who told a lie.” Instead of asking participants to determine who is lying, we asked participants to give their impressions of stimulus persons, based on the presumption that an individual positively evaluates others who tell the truth. Study 1 employed a between-participant design but we found that the instructions had no significant effects. Study 2 employed a within-participant design and we found that the lie instructions had significant effects on lie recognition in evaluating stimulus persons. In the without-lie-instruction condition, stimulus persons were more likely to be positively evaluated than those in the with-lie-instruction condition. Our study suggests that researchers should be more careful about what instructions they give to participants in lie recognition experiments.