Children learn information from a variety of sources and often remember the content but forget the source. Whereas the majority of research has focused on retrieval mechanisms for such difficulties, the present investigation examines whether the way in which sources are encoded influences future source monitoring. In Study 1, 86 children aged 3 to 8 years participated in 2 photography sessions on different days. Children were randomly assigned to either the Difference condition (they were asked to pay attention to differences between the 2 events), the Memory control condition (asked to pay attention with no reference to differences), or the No-Instruction control (no special instructions were given). One week later, during a structured interview about the photography session, the 3- to 4-year-olds in the No-Instruction condition were less accurate and responded more often with ‘do not know’ than the 7- to 8-year-olds. However, the older children in the Difference condition made more source confusions than the younger children suggesting improved memory for content but not source. In Study 2, the Difference condition was replaced by a Difference-Tag condition where details were pointed out along with their source (i.e., tagging source to content). Ninety-four children aged 3 to 8 years participated. Children in the Difference-Tag condition made fewer source-monitoring errors than children in the Control condition. The results of these 2 studies together suggest that binding processes at encoding can lead to better source discrimination of experienced events at retrieval and may underlie the rapid development of source monitoring in this age range.