It has been shown that it is surprisingly difficult to match a photograph of a face to a live face and neither humans nor machines can do this reliably. However, experimental research has allowed us to explore facial recognition and find ways to improve it. The natural image inconsistencies in lighting, expression and colour, pose problems for facial recognition from photos. Prototyped images which are created by merging several photographs of one person are found to stabilise facial appearance and have an advantage over normal photographs in tests of face recognition. It has also been suggested that when images are caricatured, which involves exaggerating spatial differences between individual faces and an average norm face, this can also improve recognition. Furthermore, one view which is introduced in this study is that images which are of a good likeness of one individual are more efficiently recognised over images which are a bad likeness. The present study involved three experiments. The first was a likeness rating task where participants rated images of familiar faces in relation to how alike they were to their own memory of what that person looks like. These ratings were used to construct stimuli for experiments two and three, which were name-verification tasks. The difference between the two tasks was the duration in which the stimulus was shown, the first was for 33ms and the second for 100ms. The results for both experiments showed that there was a main effect of likeness, the good likeness stimuli were recognised faster and more accurately than the bad likeness stimuli. There was also a main effect of caricatures for the third experiment which found that caricatures were recognised faster than non-caricatured images. Overall, no main effect of prototyped stimuli was found, yet further effects were found in the interactions between factors.