States across the country are developing systems for evaluating school principals on the basis of student achievement growth. A common approach is to hold principals accountable for the value added of their schools--that is, schools' contributions to student achievement growth. In theory, school value added can reflect not only principals' effectiveness but also other school-specific influences on student achievement growth that are outside of principals' control. In this paper, we isolate principals' effects on student achievement growth and examine the extent to which school value added captures the effects that principals persistently demonstrate. Using longitudinal data on the math and reading outcomes of fourth- through eighth-grade students in Pennsylvania, our findings indicate that school value added provides very poor information for revealing principals' persistent levels of effectiveness.