This study analyzed the effects of family educational support and family background on the change of achievement in mathematics from 5th grade to 9th grade among the students who were grouped into the underachievers at 4th grade. Even though underachievement has been considered the results mainly found among the group of disadvantaged families, poor performance in mathematics at 4th grade often occurs regardless of family background due to the rapidly rising difficulty in mathematics at that time. We assumed that family background and family educational support would become significant resources to overcome underachievement in mathematics. In the main analysis, a two-way random effect model was employed. Using the SELS (Seoul Education Longitudinal Survey), the result showed that high level of maternal education, two-parent family, and high amount of family educational support were significantly effective for improving math achievement during the period of analysis among students who underachievers at 4th grade. In addition, a high level of maternal education accelerated the effect of family educational support. Meanwhile, individual effort on math learning, in-class attitude, and peer relationship were significant in explaining the improvement of math achievement among underachievers. Particularly, since in-class attitude was more significant among the group of students whose mothers attained junior college degrees or less than the counterpart group of students, the finding showed the significance of the policy targeting underachievers with underprivileged family background.