This paper, using multi-level modelling, sought to highlight student-level and school-level characteristics that differentiate the academic performance in mathematics from a sample of 6080 Moroccan secondary students who participated in the 2011 Trends in International Mathematics and Science Study (TIMSS) test. The results indicated little evidence to support the Heyneman-Loxley effect, as they showed that individual and family characteristics were the strongest determinants of achievement compared to the schools' factors. Students scored higher when they were boys, were younger, were more self-confident, were more ambitious, and felt secure in school. They also scored higher when they came from families with a higher socioeconomic status, particularly when their parents were able to ensure the availability of learning resources, and were of a higher level of education. Finally, schools were matter for their pupil achievement mainly through the location channel, as urban pupils outperformed rural ones. However, the other school characteristics, including those related to teachers, played a negative, or at best neutral, role in pupil outcomes. These findings have important public policy implications. Redesigning education policy towards improving the school's environment and fostering inclusivity would be needed to enhance learning performance of Moroccan students.