The aim of this study is to examine the effect of self-confidence on mathematics achievement. In this meta-analysis, the effect of self-confidence on mathematics achievement was analysed using the Trends in International Mathematics and Science Study (TIMSS). In the first stage of the study, the average effect size of self-confidence on mathematics achievement was determined, and in the second stage, the moderators that may impact average effect size were investigated. For the meta-analysis, 336 independent data belonging to 76 countries included in the TIMSS (2003, 2007, 2011 and 2015) were combined, and a sample of 1,028,567 subjects was obtained. The average effect size was then calculated using the differences between means (Cohen d) based on the random effect model, whereas the significance of the moderator variables was calculated using the Q statistic. The results indicated that self-confidence has a moderate effect on mathematics achievement. In addition, the year in which the survey was conducted, national culture, continent of the country and Human Development Index were found to play a moderator role in the effect of self-confidence on mathematics achievement. In this respect, the economic, cultural and social capital of students are very similar to each other as the economic levels of the countries and individual families increase. These similarities highlight the differences between students' personal characteristics, wherein traits such as self-confidence have become one of the most important variables that determine students' mathematics achievement in recent years.