In a first experiment, adults practiced single- and two-digit mental addition over a 6-day period. There was a clear training effect for both types of problems, even if two-digit additions were different from one day to another. Moreover, participants were tested on their written calculation abilities before and after the training programme. We showed that participants who entered the mental arithmetic training
programme did not progress more in written arithmetic than participants who did not receive any training between the pre- and the post-tests. Conversely, in a second experiment, participants were trained in multidigit written addition and we examined the effect of such training on single- and two-digit mental addition. Again and trivially, there was a clear effect of training on written addition, but, more
importantly, a transfer on mental addition. The implications of these results on the nature of the relationship between mental and written arithmetic are discussed.