Most eyewitness identification research simulates single perpetrator crimes, but real-world crimes often transpire at the hands of multiple perpetrators. It is unclear how multiple perpetrators might impact the ability of eyewitnesses to discriminate between the guilty and innocent. To address this issue, we conducted two experiments in which large nationwide samples of participants read a vignette about a crime being committed, and then viewed one to two target faces. If there were two targets, they were either similar or dissimilar, and presented either simultaneously or sequentially. Participants later viewed a target-present or -absent lineup for each target. As predicted, presenting one target enhanced discriminability compared to two targets presented simultaneously (Experiments 1 and 2), and two targets presented sequentially (Experiment 2). Results were mixed regarding the effect of similarity, with discriminability better for dissimilar compared to similar target pairs in just one experiment. Lastly, high confidence indicated high accuracy across all conditions.