<p>In this paper, we investigate how labelling easy tasks with different difficulty levels influences girls’ and boys’ self-efficacy and performance in mathematics. A total of 172 girls and 177 boys from Norwegian lower secondary schools participated in an experimental study. Although all the tasks in the online survey were of equal difficulty, the students in the control group received tasks without labels, while the experimental group was assigned the same tasks labelled “easy,” “medium,” or “difficult.” Results from a Wilcoxon test showed that labelling easy tasks as difficult significantly reduced girls’ self-efficacy (with an overall effect size of 0.63), and a chi-squared test revealed a corresponding decline in their performance. In contrast, boys’ self-efficacy and performance were not influenced by the task labelling. These findings suggest that the “difficult” label should be used with caution in lower secondary mathematics education. Further research is needed to clarify why such labelling affects girls but not boys.</p>