Measures of teachers’ “value added” to students’ current test performance feature prominently in ongoing reforms to teacher evaluation systems. However, this immediate effect may not capture teachers’ more meaningful longer-term impact on student learning. Set in New York City, this study investigates the persistence of teachers’ effects. Two findings emerge. First, a teacher’s effect on students’ English Language Arts achievement has substantial crossover effects on long-term math performance. Second, the persistence of teachers’ value-added is considerably lower in schools that serve more low-achieving students or that hire fewer academically skilled teachers. The results provide evidence that teachers’ effects on students’ long-term skills can vary substantially as a function of both instructional content and quality.