We investigate the effects of private schools on reading and numeracy scores using rich population data. Conditional on lagged test scores and narrowly defined neighborhood indicators, Catholic and non-Christian faith private schools on average raise test scores by 0.18 standard deviation or more relative to the average public school, while non-Catholic Christian private schools have negligible effects. The effects of secular private "prep" schools are similar to those of Catholic schools, but selection bias is a greater concern in this case. We use school-specific estimates of effectiveness to investigate private school choice decisions and the determinants of private school effectiveness.