Abstract This study tested the theoretical proposition that the developmental increase in memory span results from a corresponding increase in the use of grouping strategies. With slow presentation, experimenter-provided grouping eliminated the span differences between sixth graders and adults but, while grouping helped first graders more than adults, a sizable difference in digit span remained. With fast presentation, grouping increased the adults' performance more than the childrens'. Another finding was that both groups of children had greater digit spans with fast presentation than with slow. For adults, the reverse was true.