The verbal overshadowing effect is the phenomenon in which describing a previously seen face impairs its recognition (Schooler and Engstler-Schooler, 1990). The primary purpose of this research was to investigate how the similarity between a target and distractors influences verbal overshadowing. In order to manipulate test-set similarity, we blended the faces of different people by using morphing techniques. As a result, verbal overshadowing was found when test-set similarity was relatively high, while the effect was not evident when there was a lesser degree of similarity. The results suggest that replicating the emergence of the verbal overshadowing effect depends on test-set similarity. The implications of these findings for research and practice are discussed.In criminal investigations, obtaining a thorough description of the perpetrator of a crime is often crucial to the investigation, because it is believed to assist officers in quickly identifying and apprehending a suspect . Do such descriptions really assist in the identification of a suspect? Recent research suggests, on the contrary, that such descriptions can disrupt subsequent identification performance. For example, Schooler and Engstler-Schooler (1990) initially showed participants a video of a bank robbery for 30 s, and subsequently directed half of the participants to describe the appearance of the robber and the other half to complete a filler activity. All the participants were then given a recognition test that included photos of the target face and seven distractor faces of similar appearance. The participants who had described the face were less accurate in choosing the target face than the participants who had not described it; that is, a verbal description impaired the subsequent identification of a target face. They termed this phenomenon verbal overshadowing.Since Schooler and Engstler-Schooler (1990) discovered that verbalization disrupted subsequent face recognition, a number of studies have replicated their results (Ryan and Schooler,