The No Child Left Behind Act imposes sanctions on schools if the fraction of each of five racial groups of students demonstrating proficiency on a high stakes exam falls below a statewide pass rate. This system places pressure on school administrators to redirect educational resources from groups of students most likely to demonstrate proficiency towards those who are marginally below proficient. Using statewide observations of 3rd and 4th grade math tests, this paper demonstrates that students of successful racial groups at schools likely to be sanctioned gain less academically over their subsequent test year than comparable peers at passing schools. This effect is stronger at schools more likely to suffer from NCLB sanctions and is robust to nonrandom sample selection. (Contains 6 tables, 13 notes, and a bibliography.) [Paper presented at the NCLB: Emerging Findings Research Conference at the Urban Institute, Washington, D.C., Aug 12, 2009.]