This article considers potential impediments to the estimation of teacher quality caused primarily by the purposeful behavior of families, administrators, and teachers. The discussion highlights the benefits of accounting for student and school differences through a value-added modeling approach that incorporates a student's history of family, school, and community influences. Unfortunately, the value-added framework does not address all potential impediments to the consistent estimation of teacher quality of instruction, even when focused solely on within-school differences. Of particular importance is the nonrandom assignment of students to classrooms on the basis of unobservable characteristics and parental and school responses to teacher quality. Evidence from studies that take extensive measures to mitigate the influence of purposeful classroom assignment nonetheless supports the existence of substantial within-school variation in the quality of instruction.