This paper addresses the problematic nature of pupils' attempts to use their science conceptions in contexts other than the ones in which the original learning of the concept takes place. It reports research with 60 Year-5 primary school children studying current electricity, during which the researcher employed metacognitive activities alongside normal teaching procedures, in an attempt to enhance cross-contextual use of taught concepts. In order to assess the effect of different contexts on pupils' performance, a test was repeatedly administered over one school year that tested the same concepts in distinctly different contexts. Although the role of a familiar context is only partly determined, the results suggest that pupils' ability to use concepts in unfamiliar contexts is stable for a long period of time. The results also support the provision of metacognitive activities as a means of enhancing pupils' ability for cross-contextual use of their conceptions of science.