Purpose. Children who witness crimes are sometimes asked to view a photospread lineup to see if they can identify the culprit. Here, we examined the effectiveness of two manipulations designed to assist in overcoming the tendency of child eyewitnesses to choose from lineups, a tendency that manifests in false identifications from targetabsent lineups. Methods. Children aged around 12 years (N = 432) were randomly assigned to control, Not Sure, or accuracy motivation conditions in order to examine both targetabsent and -present identification test performance across multiple sets of stimulus materials. Results. The Not Sure option did not affect false alarms or hit rates, but the combination of accuracy motivating conditions and the Not Sure option reduced guessing, with overall rates of false identifications falling by 12.2%. The latter effect was much stronger under some stimulus conditions but not detectable under others, indicating that the manipulations could improve but not impair performance. While there were beneficial effects at the group level, the manipulation did not produce a general change in individual children's decision strategies. Conclusions. Eyewitness identification test manipulations that reliably reduce false alarms, without lowering hits, by child witnesses have not yet been identified. Here, we showed how a simple-to-implement accuracy motivation manipulation reduces false alarms under some conditions, while also highlighting the importance of evaluating the efficacy of interventions across varied encoding and test conditions. However, developing a procedure that can produce a generalized improvement in decision making at the individual level remains a major challenge for eyewitness researchers.