Many experimental studies from the 70s and 80s show that children do not understand metaphors until fairly late in development (not until adolescence, some claim). This paper argues that children's metaphorical abilities may not be as weak as they first appear. Findings suggesting a poor comprehension of metaphor by young children might be better explained by factors other than purportedly inadequate pragmatic abilities. Furthermore, attested cases of metaphor production by children have often been re-analysed either as cases of overextension (i.e., erroneous extension of the term's conventional denotation) or as cases of pretence, and are thus not considered to be genuine metaphors. This paper explores the hypothesis that such re-analyses do not preclude the possibility that young children possess the necessary abilities to produce metaphors. Instead, some aspects of overextension and pretence may pave the way to metaphorical abilities.