This paper examines the effects of policies that increase the number of students who take the first course in algebra in 8th grade, rather than waiting until 9th grade. Extending previous research that focused on the Charlotte-Mecklenberg school system, we use data for the 10 largest districts in North Carolina. We identify the effects of accelerating the timetable for taking algebra by using data on multiple cohorts grouped by decile of prior achievement and exploiting the fact that policy-induced shifts in the timing of algebra occur at different times in different districts to different deciles of students. The expanded data make it possible to examine heterogeneity across students in the effect of taking algebra early.  We find negative effects among students in the bottom 60% of the prior achievement distribution. In addition, we find other sources of heterogeneity in effects.