We use panel data in Washington State to study the extent to which teacher assignments between fourth and eighth grade explain gaps between advantaged and disadvantaged students--as defined by underrepresented minority status (URM) and eligibility for free or reduced price lunch (FRL)--in their eighth grade math test scores and high school course taking. We find some significant gaps between advantaged and disadvantaged students in the value added of the teachers to which they are assigned in these grades, although gaps in middle school grades are sensitive to the specification of value added. We then show that teacher assignments are highly predictive of both eighth-grade test scores and advanced course taking in high school, and that differences between advantaged and disadvantaged students in teacher assignments explain significant portions of student outcome gaps. In the case of eighth-grade test scores, the URM/non-URM gap drops by more than 15% and the FRL/non-FRL gap drops by more than 20% when we control directly for teacher assignments. That said, the reduction in the achievement gap is more modest when we control for measures of teacher value added that do not control for classroom characteristics (8% and 9%, respectively), while gaps actually increase slightly when we control for measures of value added that control for classroom characteristics. These patterns are qualitatively similar and even larger in magnitude when we consider the number of advanced math courses taken in high school as the outcome.