This study used large-scale regional monitoring data of eighth-grade mathematics students at the compulsory education stage from various areas of mainland China. It extracted a total of 156,661 students and 4,676 junior high school mathematics teachers from 146 districts and counties located in six regions (provinces or cities). The study analysed academic achievement in mathematics in these areas and established a hierarchical linear model to explore the factors affecting academic achievement at different levels. The results are as follows: (1) approximately 94% of eighth-grade students reached the level C academic benchmark - students in East China had the highest compliance rate with this level, followed by those in North, South, and Central China; (2) girls, non-leftover students, and children without siblings performed better, and urban students performed significantly better than county and rural students; (3) approximately 34% of students' mathematics academic performance came from inter-school variability - regional background had a greater impact on mathematics than did teaching factors, while urban and rural background had the least impact. In contrast, the influence of individual characteristic variables was higher than that of student background variables, including a greater positive effect of self-efficacy and a greater negative effect of mathematics anxiety.