During a premeditated crime, perpetrators often wear a disguise such as a ski mask to hinder subsequent identification. Ski masks occlude facial features, which impedes holistic processing—an important component of face recognition (Tanaka & Simonyi, 2016). Despite significant advances in eyewitness identification research, there are no evidence-based recommendations for lineup construction for cases involving a masked perpetrator. The purpose of this dissertation was to examine identification accuracy of a masked perpetrator as a function of lineup type (i.e., full-face or masked-face lineups) and target presence (i.e., target-absent or target-present). Participants completed four trials (one per condition). For each trial, participants watched a mock crime featuring a masked perpetrator then were administered a lineup. I predicted superior identification performance from masked-face lineups compared to full face lineups because masked-face lineups provide a perceptual match to a masked perpetrator whereas a full-face lineup does not. The aim of Experiment 2 was to conceptually replicate Experiment 1 with additional measurements. Participants were asked to indicate their lineup preference then predict their ability to make an accurate lineup decision from either a masked-face lineup or a full-face lineup (pre-ID confidence). In both Experiments 1 and 2, identification of a masked perpetrator was superior when participants were administered a masked-face lineup compared to a full-face lineup. Moreover, participants showed no preference and comparable pre-ID confidence for either lineup on the first trial. However, preference and pre-ID confidence ratings shifted to favor masked-face lineups on subsequent trials. Experiment 3 examined how mock jurors perceive an identification made from a masked-face lineup compared to a full-face lineup. Crucially, mock jurors did not discount masked-face lineup identifications compared to full-face lineup identifications. My data suggest that presenting a lineup containing faces that better match the perceptual appearance of the originally encoded target can enhance identification performance.