The main purpose of the present study was to examine the long-term effects of a previous quasi-experiment that used inner speech as a tool to promote appropriate strategies in multiplication. Results one year after the intervention were published by Ostad and Askeland. This is a follow-up study. The 50-week intervention program was conducted in third and fourth grade and was modelled in accordance to the developmental course of private speech (from audible private speech to silent inner speech). Private speech and task-specific strategy use were examined in the fourth and seventh grade pupils in order to evaluate whether different teaching programmes resulted in private speech and task-specific strategy use differences three years after intervention. While internalisation of private speech appeared to stagnate from the fourth to seventh grade in the I-group, the results from more internalised task-specific strategy use indicated that the intervention programme was successful. This study supports the hypothesis that internalisation of inner speech at an early stage of development may be associated with more appropriate strategy use, both short and long term. (Contains 3 tables and 13 figures.)