There is evidence that inducing a luck-related superstition leads to better performance on a variety of motor dexterity and cognitive tasks. However, some replication efforts have failed to succeed. At the same time, our previous findings suggest that the effect of good luck belief on cognitive performance interacts with gender. The present research aimed at replicating the study with a sample of adolescents among whom the superstitious beliefs are particularly prevalent. The participants (N = 99) were allocated to either a control or experimental group, and were asked to solve eight problems focused on cognitive reflection, conjunction fallacy, denominator neglect, and probabilistic reasoning. The experimental manipulation negatively affected boys' performance. Yet, it facilitated performance in girls via increase in their self-efficacy, measured as subjective estimate of future success in the tasks. Thus, gender seems to moderate the effect of luck-related belief on solutions to cognitive problems, which are an important part of our day-to-day decisions. Given initial gender gap in the present tasks, the crucial question to be addressed in future research is possibility of gender being a proxy for prior competence. It would imply that good luck beliefs might help low scorers, for instance in becoming less anxious and more confident, but could be harmful for high scorers.