We estimate the effect of school size on students' long-term outcomes such as high school completion, being out of the labor market, and earnings at the age of 30. We use rich register data on the entire population of Danish children attending grade 9 in the period 1986-2004. This allows us to compare the results of different fixed effect and instrumental variables estimators. We use the natural population variation in the residential catchment areas and school openings and closures to instrument for actual school size. We find a robust positive but numerically fairly small relationship between school size and alternative measures of long-term success in the educational system and the labor market. The positive impact of school size seems mainly to be driven by boys, students from families with a low educational level and students attending schools in urban areas.