Background: Personality is reliably associated with academic performance, but personality measurement in primary education can be problematic. Young children find it difficult to accurately self-rate personality, and dominant models of adult personality may be inappropriate for children. Aims: This meta-analysis was conducted to determine the validity of the Five-Factor Model (FFM) of personality for statistically predicting children's academic performance. Sample: Literature search identified 12 reports, with cumulative sample sizes ranging from 4,382 (19 correlations) to 5,706 (23 correlations) for correlations with Emotional Stability and Conscientiousness respectively. Method: Hunter-Schmidt random-effects meta-analysis was used, and moderators were tested using sample-weighted regression. Results: When compared with self-rated measures, adult-rated Conscientiousness and Openness were more strongly correlated with academic performance, but adult-rated Agreeableness was less strongly correlated. Q-set-based assessments had lower validity, which appeared to explain moderating effects of rating source. Moderating effects were not found for age, year of education (grades 1-7), or language within which the study was conducted. Conclusions: Conscientiousness and Openness had two of the strongest correlations with academic performance yet reported, comparable with previous meta-analytic correlations of academic performance with instructional quality, cognitive ability, and feedback. The FFM appears to be valid for educational research with children. Openness, which has no counterpart in models of children's temperament, should be further researched with children. Future research should examine the measurement of childhood personality, its relationship with intelligence, the extent to which it is malleable in primary education, and its causal relationship with academic performance.