To date, research investigating the similarity among lineup members has focused on adult eyewitnesses. In the present research, children made identifications from lineups containing members of lower or higher similarity to a target person. In Experiment 1, following a live interaction, children's (6–14 years) correct identification rate was reduced in higher-similarity relative to lower-similarity lineups. In Experiment 2, children (6–12 years) and adults watched a video containing a target person. Again, higher-similarity lineup members reduced children's correct identifications; however, similarity had no effect on adults' correct identification rate. Although children benefited from lower-similarity lineups when the target was present, lower-similarity lineups generally increased misidentifications of an innocent suspect when the target was absent. Thus, increasing similarity in lineups for children had a cost on target-present lineups and a benefit on target-absent lineups. Copyright © 2014 John Wiley & Sons, Ltd.