Background: Math anxiety (MA) and math achievement are generally negatively associated. Aims: This study investigated whether and how classroom engagement behaviors mediate the negative association between MA and math achievement. Sample: Data were drawn from an ongoing longitudinal study that examines the roles of affective factors in math learning. Participants consisted of 207 students from 4th through 6th grade (50% female). Methods: Math anxiety was measured by self-report using the Mathematics Anxiety Scale for Children (Chiu & Henry, 1990, Measurement and valuation in Counseling and Development, 23, 121). Students self-reported their engagement in math classrooms using a modified version of the Math and Science Engagement Scale (Wang et al., 2016, Learning and Instruction, 43, 16). Math achievement was assessed using the Applied Problem, Calculations, and Number Matrices subtests from the Woodcock-Johnson IV Tests of Achievement (Schrank et al., 2014, Woodcock-Johnson IV Tests of Achievement. Riverside). Mediation analyses were conducted to examine the mediating role of classroom engagement in the association between MA and math achievement. Results: Students with higher MA demonstrated less cognitive-behavioral and emotional engagement compared to students with lower MA. Achievement differences among students with various levels of MA were partly accounted for by their cognitive-behavioral engagement in the math classroom. Conclusions: Overall, students with high MA exhibit avoidance patterns in everyday learning, which may act as a potential mechanism for explaining why high MA students underperform their low MA peers.