The current study developed a new paradigm to determine the age at which children begin to show the self-reference advantage in memory. Four-, 5-, and 10-year-olds studied lists of colourful object pictures presented together with self or other face image, and participants were asked to report aloud "who is pointing at the (object)." Then incidental free recall was carried out, followed by source judgments based on the earlier test where participants had to distinguish who pointed to the object. In Experiment 1, only 5-year-old children showed self-reference advantage in the recall, but not in source judgments. By increasing task demand in Experiment 2, 5 and 10-year-olds also showed the self-reference advantage in the recall, but not in source judgments. These results indicated that the new paradigm is appropriate to measure children's self-reference effect in memory, and children as young as 5 years begin to show this effect. (Contains 2 tables, 2 figures, and 1 footnote.)