The TIMSS is an international large scale assessment study promoted by the International Association for Educational Achievement. The study initiated in 1995 and occurs every four years since. In the 2019 edition 58 countries and educational systems took part. In this paper we used a 3-level HLM to identify which student and family (level 1), teachers and schools (level 2) and country (level 3) variables predict mathematics achievement at grade 4. At the student and family level home resources for learning (e.g., books at home), students’ self-concept in mathematics and parents’ expectations and early literacy tasks were both significant and relevant (effect size larger than 0.1) predictors of student achievement. At the school and teacher level, both percentage of disadvantage students, library books, as well as teachers’ perceptions of parents’ engagement were significant and relevant predictors. Finally, at the country level, life expectancy (a proxy of country development), GDP and teacher to pupil rations were the most significant predictors of Whitin country achievement. Countries’ variance accounted for 43% of the variance of TIMSS math scores at grade four, whilst schools’ variation accounted for 19% of this variance.

--