This study examined the interaction effect between two concept requirements (without and with concept application) and two task contexts (everyday and scientific) on pupils' responses to questions about control of variables, some of which were previously assessed in the UK national science monitoring (APU). Three hundred and twenty four pupils in Seoul, Korea, of both sexes and in three age groups (12,13 and 16) were given altogether twelve pairs of paper and pencil questions, six without concept application and another six with concept application. Results revealed that there was a significant interaction between concept requirements and task contexts, showing that pupils' achievements in without conceptapplication questions were significantly higher in everyday contexts than in scientific contexts, whereas in with concept-application questions there was no statistical difference between the two contexts. Effects of age and gender and correlations with school achievements are also discussed.