Data from 30 studies of the magnitude of gender differences in science achievement previously examined in two separate reviews were synthesized using modern methods for meta-analysis. Two meta-analysis methodologies were used: analyses of effect sizes (standardized mean differences) and vote-counting estimation procedures. Analyses revealed that magnitudes of gender differences in science achievement varied according to the subject matter under study. Males showed significant advantages in studies of biology, general science, and physics, but significant differences were not found for studies of mixed science content, and geology and earth sciences, or in a single study of chemistry. However, in all cases the numbers of effects examining each subject-matter area were quite small (seven or less). Also, studies which had focused on gender differences evidenced larger gender differences than other studies. Other possible predictors of the magnitudes of gender differences, such as grade level of the subjects and test length, did not account for significant amounts of variation among these study outcomes. Consequently, they also did not provide as strong an explanation of the patterns of gender differences as did subject matter and focus on gender.