We examine how different measures of teacher quality are related to students' long-run trajectories. Comparing teachers' "test-based" value-added to "nontest" value-added -- based on contributions to student absences and grades -- we find that test and nontest value-added have similar effects on the average quality of colleges that students attend. However, test-based teacher quality measures have more explanatory power for outcomes relevant for students at the top of the achievement distribution such as attending a more selective college, while nontest measures have more explanatory power for whether students graduate from high school and enroll in college at all.