It has frequently been demonstrated that repeated presentation of a stimulus can result in facilitated processing of the item, an effect termed repetition priming. Questions remain, however, regarding the boundary conditions of this effect, particularly for faces. For example, is repetition priming for unfamiliar faces dependent on the presentation of identical stimuli at study and test? This question was explored in three experiments in which the pose (i.e., frontal vs. ¾) and perceptual distance from the original facial identity (i.e., 100%, 75%, 50%, or 25% of original person) were manipulated between the testing phases of a standard repetition-priming paradigm. The results revealed that priming did not persist following any change to a face between study and test, thereby suggesting that repetition priming for unfamiliar faces is form specific. The theoretical implications of this finding are considered.