The theory of action behind the No Child Left Behind Act of 2001 is that "shining a light" on subgroup performance will increase reading and math test scores for minority and disadvantaged students. Using a panel of all students in Grades 3 through 8 in North Carolina from 2000 to 2008 (N = 1.7 million students in 1,800 schools), the authors estimate double- and triple-differenced models with school fixed effects to examine whether subgroup-specific accountability threats increase high-stakes test scores. These sanctions are found to have positive effects for minority and disadvantaged students. Larger positive effects emerge for the lowest achieving schools rather than schools near the margin of passing. Some evidence of adverse effects is also found for low and high achievers in math, but not in reading, a finding attributed to the combination of increases in the rigor of state standards in math and responses to an accountability metric based on test score status rather than growth. The implications of the findings for the design of educational accountability systems are discussed. (Contains 2 figures, 5 tables, and 15 notes.)