This article addresses the fate of languages acquired during childhood which have not been used in a long time: are they simply lost, overridden by other languages acquired later, or are they maintained despite the lack of use? Two German subjects are tested for their knowledge of Dutch, which they acquired as a second language in their childhood. In leaning on methodology from memory research, we assess different levels of 'knowing'. As part of these tests, this study evaluates the so-called 'savings'-approach which measures subthreshold memory items which are not recalled or recognized using other types of tests. The results suggest that even after 30 years of non-use, residual vocabulary knowledge can be found. There is a strong effect of cross-linguistic influences, but also Dutch words that are not cognates seem to be retained. However, the part-set hypothesis, i.e. the reactivation of a whole set of words after the relearning of some, could not be confirmed. This study demonstrates that it can be difficult to distinguish between 'real' memory and well-developed learning strategies.