This study explored children's episodic memory. 43 7-year-olds and 43 5-year-olds took part in a contrived interaction with a "magician".After seeing a magic show,the children were interviewed immediately and assignedrandomly in three conditions:no leading(free recall), leading,and misleading. Older children recall more accurate information than younger children, and both group recall more accurate information after short delay than the long delay.No matter younger or older children, they had ability to provide accurate testi-mony.Leading question facilitated free recall after three days. Leading group were more accurate than no leading and misleading group in the misleading suggestive questions.