Predictions of the Identical Elements (IE) model of arithmetic fact representation (Rickard, 2005; Rickard & Bourne, 1996) about transfer between arithmetic facts were tested in primary school children. The aim of the study was to test whether the IE model, constructed to explain adult performance, also applies to children. The IE model predicts no positive transfer when the numerical elements of a test problem do not match exactly with those of a practice problem. In two experiments, children practiced addition or multiplication problems. A large positive transfer effect was found for problems with an operand order change; improvement was just as high as for practiced problems. Smaller transfer effects were found for problems where one operand was increased with one unit and for problems with operands more different from practice problems. Analogous results were found for addition and multiplication, suggesting that storage and retrieval processes in both domains are similar in children. The findings suggest that there is a gradual decrease in transfer as the test problems become more different from practice problems, indicating that a more sophisticated model than the IE model is needed to explain transfer effects in children's recall of arithmetic facts. (Contains 2 tables, 2 figures, and 3 footnotes.)