Nearly 5,000 self-generated science-related K-12 students' questions, classified into seven science subjects, were used to quantitatively measure the gender gap in science interests and its change with age. In this data set, a difference between boys' and girls' science interests did not exist during early childhood, but increased over 20-fold by the end of high school. Furthermore, the gap widened in a stereotypical manner, with girls being increasingly interested in biology and boys more interested in physics and technology. This method could be applied for identifying and comparing the gender gap in science interests between different populations based on different data sources.