The capacity of working memory is limited and undergoes important developmental changes during childhood. One proposed reason for the expansion of working memory capacity during childhood is the emergence and increased efficiency of active maintenance mechanisms, such as that of refreshing. Refreshing is a proposed mechanism to keep information active in working memory by bringing memory items back into the focus of attention. One prevalent view is that the spontaneous use of refreshing emerges around the age of 7, and becomes more efficient during middle childhood and beyond. Using a novel approach to examine refreshing in children in Experiment 1, we show, against common conceptions, that simply giving free time in a basic working memory task does not lead to spontaneous refreshing in 9 to 12 year old children. Instead, their focus of attention appears to linger on the last-presented memory item, even when ample time for refreshing is provided. Experimentally imposing the use of refreshing in Experiment 2, however, showed that children in this age range are able to switch their focus of attention away from the last-presented item in order to switch to another memory item. Thus, the current study uncovers that children in middle childhood do not always spontaneously switch attention away from the last-presented memory item to refresh the entire list, even though they are able to switch attention away from the last-presented memory item when instructed to do so. The theoretical implications of these findings are discussed.