Standardized testing has exploded into nearly every grade level and subject area in public education over the past two decades. Attaching high stakes to standardized test results was intended to improve education for all students, but especially for those who belong to consistently low-performing groups. Theory and experience, however, show that privileged groups maintain their advantages even in the face of education reforms. Tracking practices in particular have the potential to worsen the inequalities associated with high-stakes testing. This study uses a unique longitudinal dataset to observe the existence, growth, and harm of achievement gaps through high-stakes testing in North Carolina. The study demonstrates that high-stakes standardized tests predict college performance for students whose high school experiences were in in the top academic tracks, but not for students in lower academic tracks. Findings suggest that standardized test scores of lower track students reflect less learning of the kinds of substance and higher order thinking skills needed to excel in college.