This study aimed at examining whether the computational procedure with abacus would transfer to paper and pencil computation when third grade children had the experience to solve the same problems alternately by abacus and written computational procedures. These third graders were able to solve the problems correctly by abacus, but not always in written computation. The results showed that a transfer between the two computational procedures did not occur easily. It seemed to be due to the following: 1) The two computational procedures were located in different learning contexts and were not necessarily recognized as the same, and 2) Those subjects did not understand the meaning of each step of abacus operation. For example, they failed to recognize the equivalence between the original procedure and the reverse ordered processing on abacus (from units to tens and to hundreds) that could easily be matched with the pape r-andpencil computation.