The present meta-analysis investigated the influence of age on face recognition. A total of 19 studies with 79 comparisons of younger and older participants were included. Analyses revealed small to moderate effects for hits, and large effects for false alarms and signal detection theory (SDT) measures. Younger participants outperformed older participants on most face recognition measures. Younger participants made more hits (gu = 0.31) and fewer false alarms (gu = 0.95), thus had better SDT recognition performance (gu = 1.01) than older participants. These effects were largest for young faces, smaller for mixed-age faces, and smallest for older faces. Furthermore, older participants used a more liberal response criterion, that is, they were more likely to choose a face than younger participants (gu = 0.54). Meta-regression analyses revealed that young faces (vs. mixed-age faces) and longer retention intervals were associated with greater differences between the age groups for hits but not for false alarms. Funnel plot and trim-and-fill analyses indicated the presence of a publication bias. Theoretical implications for future research and for older people as eyewitnesses are outlined.