Gender differences in math-related professional achievements have been identified as a worldwide problem. Academic achievement assessments, however, have repeatedly revealed gender similarities. The observed gender similarity might be due to biased assessments that heavily rely on reading skills, which favors girls. The current study analyzed 29 international and within-country datasets representing a total of 9,471,692 students from 1,456 regions through four typical, large-scale student academic achievement assessments. The results showed a gender difference in mathematics achievements of greater than 0.76 (Cohen's d), favoring boys for each dataset after controlling for general reading achievements. The gender difference in mathematics achievements favoring boys exceeded 0.35 in each region, with a mean of 0.70 for 79 countries or jurisdictions in the 2018 Programme for International Student Assessment (PISA 2018) after controlling for general reading achievements. Dataset- and region-level gender differences are robust, suggesting that there is a clear gender difference in mathematics achievements that previous analyses have not identified due to the effect of reading achievements differences.