In the present study, we examined recognition performance with back view of a person in an accidental recognition paradigm. Participants performed a dummy evaluation task of individually presented photographs of persons (either front view or back view of male or female individuals) without knowing that they would subsequently perform an accidental recognition test with front view. The recognition performance was generally low when the back view faces were shown during the evaluation session, compared with when the front view faces were shown. However, it was better than chance level except when male participants viewed female photographs. The results also supported the female advantage in person recognition (i.e., female participants were generally better) and partly the own gender bias even with the back view photographs (i.e., person recognition of persons in the same gander group was better). One possible explanation for the present findings would be that the recognition of persons with memory of back-views is based on the outlines and other contextual information (e.g., head-shape, hair-style, hair color) and the female participants might pay more attention to those contextual information, which might lead to the higher recognition performance.