The effect of school inputs upon educational outcomes has been extensively studies in the literature but there are still no conclusions as to its impacts. This paper presents results that suggest a strong impact of inputs upon age-grade lag for Brazilian primary schools between 1998 and 2005. We use more than one estimation method, but the results we believe are those from a fixed effects model based upon a panel of schools. The data come from the annual School Census from 1998 to 2005 which provides information on a cohort of 132.603 schools; municipal data come from the 2000 Demographic Census. Our most important result is that school inputs have strong positive effects upon educational outcomes, particularly in the case of those schools whose inputs are especially low.