This paper identifies the amount of variance in mathematics achievement in high- and low-achieving schools that can be explained by school-level factors, while controlling for student-level factors. The data were obtained from 2679 Iranian eighth graders who participated in the 2007 Trends in International Mathematics and Science Study. Of the total sample, 1422 and 1257 students were from high- and low-achieving schools, respectively. Two-level hierarchical linear modelling was applied. The results indicated that of the total variance in mathematics achievement, 27.95 and 6.70% were due to between-school differences in the high- and low-achieving schools, respectively. Controlling for the school-level factors, the better-performing students were those with a higher level of confidence in learning mathematics in both samples. After controlling for the student-level factors, inadequacies in school resources and school type yielded the strongest link to achievement in the high- and low-achieving schools, respectively.