School districts are a focus of education reform efforts in the United States, but there is very little existing research about how important they are to student achievement. We fill this gap in the literature using 10 years of student-level, statewide data on fourth- and fifth-grade students in Florida and North Carolina. A variance decomposition analysis based on hierarchical linear models indicates that districts account for only a small share (1 to 2 percent) of the total variation in student achievement. Nevertheless, the differences between lower- and higher-performing districts are large enough to be of practical and policy significance, with a one standard deviation difference in district effectiveness corresponding to about 0.11 standard deviations in student achievement (about nine weeks of schooling). District performance is generally stable over time, but there are examples of districts that have shown significant increases or decreases in performance.