Eighty nursery school children were randomly divided into four groups of 20 and given a serial short-term memory task in which difficult-to-label stimuli were used. Three experimental groups were provided with labels for the stimuli. Of these, one group overtly pronounced the labels and rehearsed them during the task, one group merely pronounced the labels overtly, and one group was instructed to say the labels covertly. A control group received no labels for the stimuli. Rehearsal of the labels was found to facilitate recall performance on early serial items, and overt labeling facilitated recall on the last serial item. Covert labeling did not facilitate recall. The results supported the hypothesis that qualitatively different processing strategies determine primacy and recency effects. Current theories of the role of verbalizing in children's memory performance are discussed. (Author/MS)