This study investigates infants' (ages 15-28 months) performance on two visual self-recognition measures: using a videoscreen as a mirror-like medium to locate objects not seen directly, and locating a mark on the face after seeing it in the mirror-like medium. Infants experienced one of two training conditions on the video-screen: they watched their parents and the parents' reflected images or they watched themselves. The study aims to find evidence for the view that object-locating and mark-directed behaviour in self-image presentation are in different ways experience-dependent. Contingency experience influenced the ability to locate objects, but did not influence mark-directed behaviour in a marked-face (rouge) condition. Also, children with a higher level of object-permanence understanding scored better on object-localization, while level of object-permanence understanding was unrelated to mark-directed behaviour in the rouge condition. The findings suggest that mark-and object-locating conditions follow a different developmental course. Search behaviour in this paradigm is also discussed with regard to experiential elements, perceptual cues and representational skills. The locating of mirrored objects may be a good paradigm for studying sensorimotor information processing, without being necessarily used as an indicator of self-awareness. Copyright © 2001 John Wiley & Sons, Ltd.Key words: mirror training; object-permanence; self-awareness; self-recognition; spatial knowledge Visual self-recognition of one's mirror image, which may mark the beginning of conceptual self-awareness in children, has become a standard paradigm for assessing development of self-awareness in the preverbal period. In the mirrorand-rouge or mark technique, a spot of rouge is surreptitiously applied to the nose or the cheek of the infant, to elicit self-directed behaviour when looking at