We estimate the impact of school characteristics on house prices using data from Chicago for 1987-1991. We merge information from the American Housing Survey and the Illinois School-Report Cards and assign to each house the school-level data for the closest school. The evidence suggests that the school-level variables are significantly better in describing house values than the district-level data. We find that controlling for unobserved, temporally-stable determinants of house values is necessary to obtain unbiased estimates of the impact of school characteristics on house prices. Using this specification, we find that homeowners pay attention to school outputs, i.e., test scores (the elasticity is approximately one), and not to inputs, i.e., per-pupil expenditures. (C) 2002 Elsevier Science (USA). All rights reserved.