Many studies of simulated eyewitness situations have shown that under certain laboratory conditions, people's confidence about their identifications predicts their accuracy, but that their reported confidence can be affected by telling them that they chose the suspect. In this study, eyewitnesses (n= 134) to real crimes took part in lineups at an identification suite in the United Kingdom and were asked questions about their memory both before and after they were told whether they had identified the suspect or a filler. Before the eyewitnesses were told whether they had identified the suspect or a filler, their responses to several questions reliably differentiated between those who identified the suspect and those who identified a filler. In addition, responses to the memory questions were affected by telling the eyewitnesses whether or not they had identified the suspect. These results show that postidentification feedback affects real eyewitnesses and highlight the importance of recording meta-memory variables before an eyewitness discovers whether he or she has identified the suspect.